JPH10134120A - Method and device for table processing - Google Patents

Abstract

PROBLEM TO BE SOLVED: To easily collate a character recognition result with a table image on a display screen. SOLUTION: A table image storage means 12 stores the image of a table composed of an item frame and a data frame in which entry contents belonging to item names entered into the item frame are entered. A ruled-line extracting means 14 extracts ruled lines out of a read image and an in-table frame extracting means 16 extracts an in-frame table encircled with the ruled lines. An item frame extracting means 18 extracts the item frame out of the extracted in-table frame and a data frame extracting means 20 extracts the data frame. A character recognizing means 22 recognizes the characters of the item name and entry contents entered into the extracted item frame and data frame respectively. In a display table 26, the display format of the data frame is set corresponding to the item name that the entry contents entered into this data frame belongs to and differently corresponding to the individual item names. A display control means 30 when displaying the table image and character recognition result retrieves the display format of the data frame corresponding to the recognition result of the item name stored in a table information storage means 24 from the display table 26 and instructs it to a display means 28.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a table processing method and a table processing apparatus for extracting a character pattern from a character frame in a table area in an image image read from a document or a form, and performing character recognition.

[0002]

2. Description of the Related Art In general, a table processing apparatus includes an image reading unit,
The image processing apparatus includes a processing unit that processes the read image by a computer, a display unit that displays a result of processing by the processing unit, and an input unit that performs input control. When processing an image obtained by reading an image of a document, form, or the like, a character area, a graphic area, a photograph area, and a table area are extracted from the image, and different processing is performed on each area. Is often applied. Among these table areas, a process of detecting a character frame (or a character column) in the extracted table region, extracting a character image from an image image in the character frame, and performing character recognition is performed. Will be applied.

In general, frames (or also referred to as columns) constituting a table include an item frame serving as a subheading and a data frame in which contents corresponding to the item frame are entered. Then, the item name described in the item frame and the entry content described in the data frame corresponding to the item frame can be recognized in association with each other (refer to the document “Transactions of the Institute of Electronics and Communication Engineers '86 / 3”. Vol.J69-D No.3 p
p400-409 "). Whether the character recognition result is erroneous or not is verified by comparing the character recognition result with the image of the table on the display screen.

[0004]

As described above, the operator checks the image name of the table area displayed on the display screen and the character recognition thereof in order to check whether the item name or the entry content is erroneously recognized. Check with the result. However, when performing the collation, it is difficult to understand where each table information such as the item name and entry content in the image image and the recognition result of the recognized item name and the recognition result of the entry content are located on the display screen, Therefore, there is a problem in that a burden is imposed on an operator who corrects the character recognition result, causing an operation error. In particular, when the display style (format) of the table differs between the image image and the character recognition result, for example, the item name of the image image corresponding to the recognition result of the item name of the character recognition result is changed to which position of the displayed image image. Is difficult to find.

Accordingly, there has been a demand for a table processing method that facilitates comparison between an image and a character recognition result, and a table processing apparatus for realizing the method.

[0006]

According to the table processing method of the first invention of the present application, an item frame, an item name described in the item frame, an entry belonging to the item name, and the entry content are described. Each table information of the data frame is converted into table image information by the image input means, and the table image information corresponding to the item name and the entry content is subjected to character recognition processing, and the result of the item name recognition and entry is performed. Acquiring the recognition result of the contents, and converting the table image information and the recognition result into the screen display information of the table display information and the recognition display information, and displaying the screen display information on the screen of the display unit. It is displayed in a display format specified by another screen display information that is physically related.

Here, the display format refers to various display methods such as coloring the frame itself, characters or the background in the frame, overlaying a mask pattern (pattern) on the frame, and underlining characters. That means.

As described above, in order to display certain screen display information in a display format specified by another screen display information, the other screen display information is used as a position reference in the display screen, so that the screen display information is displayed in the display screen. The display position of the certain screen display information described above can be easily understood.

Further, according to a preferred embodiment of the table processing method of the present invention, the image information is stored in the table information in correspondence with the image information related to the another screen display information by the certain screen display information. The display format is stored in a storage means, and the display format specified by the image information is stored in a display table memory in correspondence with the display format specified by the image information. When displaying on the screen of the means, the image information of the another screen display information is read from the table information storage means with the image information of this certain screen display information, and the display table memory is searched with the read image information. And reading the corresponding display format, and giving the read display format to the display means.

Further, according to a preferred embodiment of the table processing method of the present invention, a plurality of ruled lines are extracted from the table image information obtained by conversion by the image input means, and are surrounded by the extracted ruled lines. A plurality of table inner frames extracted from the extracted table inner frame, extracting an item frame of the table image information from the extracted table inner frame, and replacing the inner table frame adjacent to the extracted item frame with a data frame of the table image information. The image information stored in the table information storage unit is extracted by setting the item name described in the extracted item frame and the entry content described in the extracted data frame as a target of the character recognition processing. , The extracted item frame, the extracted data frame, the item name recognition result, and the entry content recognition result.

Further, in the table processing method of the present invention, preferably, the certain screen display information is used as table display information of the data frame, and the another screen display information is added to an entry described in the data frame. It is preferable to use the recognition display information of the item name to which the item belongs.

As described above, since each data frame is displayed in the display format corresponding to the item name, the position of the entry corresponding to the recognition entry can be easily understood. Therefore, it becomes easier to determine whether the character recognition processing result of the entry is correct or incorrect as compared with the related art.

In the table processing method of the present invention, preferably, the certain screen display information is set as table display information of the item frame, and the another screen display information is set to an item name described in the item frame. It is good to use the recognition display information of the entry content to which it belongs.

As described above, since each item frame is displayed in the display format corresponding to the entry contents, the position of the item name corresponding to the recognition result of the item name can be easily understood. Therefore, it becomes easier to determine whether the character recognition processing result of the item name is correct or not, as compared with the related art.

In a preferred embodiment of the table processing method according to the present invention, preferably, the certain screen display information is used as recognition display information of the entry, and the another screen display information is an item name belonging to the entry. It is good to use the recognition display information.

As described above, since the recognition result of each entry is displayed in the display format corresponding to the recognition result of the item name, the position of the recognition result of the entry corresponding to the entry becomes easy to understand. Therefore, it becomes easier to determine whether the character recognition result of the input content is correct or not, as compared with the related art.

In a preferred embodiment of the table processing method according to the present invention, preferably, the certain screen display information is used as recognition display information of the item name, and the different screen display information is
It is preferable to use the recognition display information of the entry content belonging to this item name.

As described above, since the recognition result of each item name is displayed in the display format corresponding to the recognition result of the entry content, the position of the recognition result of the item name corresponding to the item name can be easily understood. Therefore, it becomes easier to determine whether the character recognition processing result of the item name is correct or not, as compared with the related art.

In a preferred embodiment of the table processing method according to the present invention, preferably, the certain screen display information is table display information of the data frame and recognition display information of an entry content described in the data frame; Another screen display information may be used as recognition display information of the item name belonging to the entry.

As described above, since the recognition result of each data frame and the entry content is displayed in the display format corresponding to the recognition result of the item name, the position of the entry content corresponding to the recognition result of the entry content can be easily understood. Therefore, it becomes easier to determine whether the character recognition result of the input content is correct or not, as compared with the related art.

In a preferred embodiment of the table processing method according to the present invention, preferably, the certain screen display information is table display information of the item frame and recognition display information of an item name described in the item frame; Another screen display information may be used as recognition display information of the entry content belonging to the item name.

As described above, since the recognition result of each item frame and item name is displayed in the display format corresponding to the recognition result of the entry content, the position of the item name corresponding to the recognition result of the item name can be easily understood. Therefore, it becomes easier to determine whether the character recognition processing result of the item name is correct or not, as compared with the related art.

Next, according to the table processing device of the second invention of the present application, the item frame, the item name described in this item frame,
Image input means for respectively converting the table contents of the entry contents belonging to this item name and the data frame in which the entry contents are described into table image information, and the table image information corresponding to the item names and the entry contents Character recognition means for performing character recognition to obtain the recognition result of the item name and the recognition result of the entry content as recognition image information, and the table image information and each image information of the recognition image information as table display information and recognition display information. Display means for displaying the image information in place of the display information; table information storage means for storing the image information in a correspondence relationship in which the image information related to another screen display information can be read out from the certain screen display information; A display table memory for storing, for each image information, a specific display format that can be read by the image information in association with the image information; When the display information is displayed on the screen of the display means, another image information corresponding to this image information is read from the table information storage means with the image information of this certain screen display information, and the read another A display format corresponding to the other image information is output from the display table memory as image information, and an instruction is given to the display means to display the certain screen display information in the output display format. Display control means.

As described above, since one screen display information is displayed in a display format specified by another screen display information, the other screen display information is used as a position reference in the display screen. In addition, the display position of the above-mentioned certain screen display information on the display screen can be easily understood.

According to a preferred configuration example of the table processing apparatus of the present invention, the image processing apparatus further comprises a table image storage unit for storing the table image information converted by the image input unit. Extracting means for respectively extracting an item frame and a data frame as table image information corresponding to the item frame and the data frame, and extracting the table image information to be subjected to the character recognition into the extracted item frame and the extraction; Recognition of the extracted item names, the extracted data frames, and the image names stored in the table information storage means as the item names and entry contents respectively described in the extracted data frames. A result and a recognition result of the entry content are obtained.

In the table processing apparatus according to the present invention, preferably, the extracting means includes a ruled line extracting means for extracting a plurality of ruled lines from the stored table image information, and a plurality of ruled lines surrounded by the extracted ruled lines. Table frame extracting means for extracting the table frame of the above, item frame extracting means for extracting the table frame of the table information from the extracted table frame, and a table frame adjacent to the extracted item frame. It is preferable to provide a data frame extracting means for extracting the data frame.

[0027] The item frame extracting means may include a table inner frame adjacent to a ruled line whose length in a direction perpendicular to the direction of a character line of a character described in the table inner frame is longer than a reference value. It is good to adopt a configuration of extracting as a frame. Further, a configuration may be adopted in which a table frame whose length in a direction parallel to the direction of the character line of the table frame adjacent to the ruled line is within the range of the reference value is extracted as the item frame. Furthermore, a configuration may be adopted in which a table frame in which the area of the table frame adjacent to the ruled line is the area within the range of the reference value is extracted as the item frame.

In the table processing apparatus according to the present invention, it is preferable that the display format is any one of a color display, a mask pattern overlap display, an underline display, a frame surrounding line display, and the like.

In the table processing apparatus of the present invention, preferably, when the certain screen display information is used as the table display information of the data frame, the another image information is added to the entry content described in the data frame. It is good to use the table image information of the item name to which it belongs.

As described above, since each data frame is displayed in the display format corresponding to the item name, the position of the entry corresponding to the recognition result of the entry can be easily understood. Therefore, it is easy to determine whether the character recognition processing result of the entry is correct or not.

In the table processing apparatus of the present invention, preferably, when the certain screen display information is the table display information of the item frame, the another image information is replaced by an item name described in the item frame. It is preferable to use table image information of the entry contents to which the entry belongs.

As described above, since each item frame is displayed in the display format corresponding to the entry contents, the position of the item name corresponding to the recognition result of the item name can be easily understood. Therefore, it is easy to determine whether the character recognition result of the item name is correct or not.

In the table processing apparatus of the present invention, preferably, when the certain screen display information is recognition display information of a recognition result of the entry content, the another image information is
It is preferable to use the recognition image information of the recognition result of the item name associated with the recognition result of the entry.

As described above, since each recognition entry is displayed in the display format corresponding to the item name recognition result, the position of the entry recognition result corresponding to the entry is easy to understand.
Therefore, it is easy to determine whether the character recognition processing result of the entry is correct or not.

In the table processing apparatus of the present invention, preferably, when the certain screen display information is used as the recognition display information of the recognition result of the item name, the another image information is included in the recognition result of the item name. It is preferable to use the information as the recognition display information of the recognition result of the associated entry content.

As described above, since the recognition result of each item name is displayed in the display format corresponding to the recognition result of the entry content, the position of the recognition result of the item name corresponding to the item name can be easily understood. Therefore, it is easy to determine whether the character recognition result of the item name is correct or not.

In the table processing apparatus of the present invention, preferably, when the certain screen display information is the table display information of the data frame and the recognition display information of the recognition result of the entry content, the another image information is It is preferable to use the recognition image information of the recognition result of the item name associated with the recognition result of the entry.

As described above, since the recognition result of each data frame and the entry content is displayed in the display format corresponding to the recognition result of the item name, the position of the entry content corresponding to the recognition result of the entry content is easy to understand. Therefore, it is easy to determine whether the character recognition processing result of the entry is correct or not.

In the table processing apparatus according to the present invention, preferably, when the certain screen display information is table display information of the item frame and recognition display information of a recognition result of the item name, the another image information is used. It is preferable to use the recognition image information of the recognition result of the entry contents associated with the recognition result of the item name.

As described above, since the recognition result of each item frame and item name is displayed in the display format corresponding to the recognition result of the entry content, the position of the item name corresponding to the recognition result of the item name can be easily understood. Therefore, it is easy to determine whether the character recognition result of the item name is correct.

[0041]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. It should be noted that the drawings are schematically shown to the extent that the configuration, arrangement, and processing flow of the present invention can be understood, and the numerical conditions and the like described below are merely examples. There is no limitation to this embodiment.

[First Embodiment] (Table Processing Apparatus) FIG. 1 is a block diagram showing a configuration of a table processing apparatus according to an embodiment. As shown in FIG. 1, the table processing device includes an image input unit 10, a character recognition unit 22, a display unit 28, a table information storage unit 24, a display table memory 26, a display control unit 30, a control input unit Means 32;
And control means 34 for controlling between the means.

The above-described image input means 10 is means for reading table information of a table area from an information medium such as a document or a form using photoelectric conversion and converting the table information into table image information. In this embodiment, since a table composed of an item frame and a data frame in which entry contents belonging to the item name described in this item frame are described is to be processed, the item frame and the It includes the item name described in the item box, the entry contents belonging to this item name, and the data box in which the entry contents are described. Here, “frame” means “column”, and these frames, ie, columns, are defined by ruled lines. In the following description, the item frame, item name, entry content, and data frame as table information are converted into table image information, that is, image images, and displayed on the display screen as table display information based on these image images. So
The image item frame, the image item name, the image entry content, and the image data frame as the table image information, and the display item frame, the display item name, the display entry content, and the display data frame as the table display information. If confusion does not occur, the description will be given using the item frame, item name, entry contents, and data frame as they are.

The table image information is readably stored in the table image storage means 12 composed of a memory. The table processing apparatus includes an extraction unit 36 that extracts an item frame and a data frame as table image information from the table image information stored in the table image storage unit 12, respectively. The extracting means 36 includes a ruled line extracting means 14 and a table inner frame extracting means 1
6; an item frame extracting means 18; and a data frame extracting means 20. These means 12, 14, 16, 18, 2
Reference numerals 0, 22, 24, 26, 30, and 34 are configured using so-called computer hardware including a central processing unit (CPU), a memory, and an input / output device.

The ruled line extracting means 14 is provided with a table image storing means 12
Is a means for extracting ruled lines from the table image information stored in. The ruled line is extracted by detecting the positions (coordinates) of both ends of the ruled line on the image according to a conventionally known method. The table frame extracting means 16 extracts a table frame surrounded by ruled lines based on the ruled line information extracted by the ruled line extracting means 14. The table inner frame is a generic name of each frame displayed on the screen in the table display information, and is a square area inscribed in the ruled line. The extraction of the inner frame is performed by detecting the position of each vertex of the square area on the image according to a conventionally known method.

Next, the item frame extracting means 18 extracts an item frame from the table frame extracted by the table frame extracting means 16. This extraction method may be in accordance with a conventionally known method. In this embodiment, however, the item frame extraction unit 18 determines the direction (vertical direction) perpendicular to the direction (horizontal direction) of the character line of the character described in the table inner frame. ), The inner frame adjacent to the ruled line whose length is longer than the first reference value has the second length in the direction parallel to the direction of the character line.
The length within the range of the reference value and the table frame having the area within the range of the third reference value are extracted as the item frames. The first reference value, the second reference value, and the third reference value are values determined in consideration of the shape and size of the table to be processed and are determined by experience through repeated tests. Stored in memory. In this embodiment, a horizontal writing table in which the direction of a character line is a horizontal direction (main scanning direction) is to be processed. In this case, generally, in this case, the item frame is located on the right side of a ruled line having a predetermined length or more. Since the first reference value is often determined, the value of the first reference value may be determined based on the regularity. In addition, the second reference value is generally determined based on the regularity that the length of the item frame in the row direction is shorter than the length of the data frame in the row direction. For example,
As a second reference value, it is empirically preferable to use "a width of 2/3 of the vertical length of each frame". In addition, the third reference value is generally determined based on the regularity that the area of the item frame is smaller than the area of the data frame.

The data frame extracting means 20 is a means for extracting data frames belonging to each of the item frames extracted by the item frame extracting means 18 from the in-table frames. In this embodiment, a horizontal writing table in which the direction of a character line is horizontal (main scanning direction) is to be processed. Therefore, in general, a data frame is often adjacent to the right side of an item frame. The data frame extracting means 20 is configured to automatically extract a table inner frame adjacent to the right side of the item frame extracted by the item frame extracting means 18.

The character recognizing means 22 obtains the character recognition result of the item name and the entry content described in each of the extracted item box and data frame as the recognition image information of the recognition result of the recognition data frame and the entry content, respectively. Means. That is, the character image in the item frame is converted into a character code to obtain the recognition result of the item name, and the character image in the data frame is converted into the character code to obtain the recognition result of the entry.
As will be described later, since the item frame and the data frame are extracted in association with each other, it is possible to acquire the recognition result of the item name and the recognition result of the entry content in association with the entry content belonging to the item name. it can.

The table information storage means 24 stores the table image information of the extracted item frame and data frame, the recognition result of the acquired item name, and the recognition image information of the recognition result of the entry. In storing the information, the information to be stored is associated with each other. If there is a predetermined correspondence between each table display information, each piece of recognition display information, and between the table display information and the recognition display information, this correspondence is a table image on which the corresponding display information is to be displayed. This is performed between each of the information and the recognition image information. Here, the table image information and the recognition image information are collectively referred to as image information.

According to the present invention, it is possible to easily visually recognize that pieces of screen display information related to each other among pieces of individual screen display information such as each table display information and each piece of recognition display information displayed on the screen. It is characterized in that one of the two screen display information is displayed in a dedicated display format according to the other screen display information. That is, each table display information of the item frame, the item name, the entry content, and the data frame, and each of the recognition display information of the item name recognition result and the entry content recognition result are information displayed on the screen. These are collectively referred to as screen display information. Each screen display information is configured to be displayed on the screen in a display format determined by another one of the screen display information having a relation with each other, and accordingly, some screen display information is displayed on the screen. The screen display information can easily confirm which other screen display information corresponds (related) to the display information depending on the display format.

Accordingly, in the display table memory 26, certain specified screen display information is stored in the display table memory 26 together with other screen display information displayed in association with the specified screen display information that determines the display format of the specified screen display information. The display format is readably stored.

For example, assuming that the designated screen display information is a data frame, the other screen display information related to (corresponding to) the data frame, which specifies the display format of the data frame, is the entry content described in the data frame. The name of the item to which it belongs. Therefore, in the display table memory (hereinafter, also simply referred to as a display table) 26, for example, the display format of the data frame is made to correspond to the item name to which the entry content described in this data frame belongs, They are set differently depending on the situation.
For example, it stores information such as displaying a data frame in which entry contents belonging to the item name “A” are described in a display format of “a”, and recognizes image information of a recognition result of the item name of the item name “A”. Is input, the information "a" is output.

Display formats include colored display, mask pattern superimposed display, underline display, frame surrounding line display, character font display, and other formats.
A format in which the data frame is displayed in color and a format in which the data frame is displayed by overlapping a mask pattern are preferable from the viewpoint of visibility. In the case of a display format in which data frames are displayed in color, each data frame is set to a different color for each data frame.

The display means 28 is a means for displaying an image image and a character recognition result as screen display information of table display information and recognition display information, respectively. The image image, that is, the table image information, is stored in the table image storage unit 12. The character recognition result is stored in the table information storage unit 24 as recognized image information. These are displayed using, for example, a display (display) device. As described above, according to the present invention, both the image image and the character recognition result are displayed as the table display information and the recognition display information, and these are compared with each other to confirm whether or not the character recognition result has an error.

The display control means 30 displays the table information storage means 2 when displaying the image on the display screen.
4, one of the two pieces of display information having a correspondence relationship with each other, for example, the other display information corresponding to the recognition result of the item name (for example, a data frame display format).
From the display table 26 and instruct the display means 28. For example, the recognition result of the item name "A" is read from the table information storage means 24, and the item name "A" is read.
Display format "a" of the data frame corresponding to
6 and instruct the display means 28. Then, the display unit 28 displays the data frame corresponding to the item name “A” in the display format “a”.

The means 12, 14, 16, 18, 20, 22, 24, 26, 2 constituting the table processing apparatus described above.
The control of the control means 8 and 30 is performed with the help of the control input means (keyboard, mouse, etc.) 32 as necessary.
4 is performed.

As described above, certain display information on the display screen, for example, the display format of the data frame is displayed according to the other display information corresponding to the display information, for example, the item name. The entry to be made can be easily found with the naked eye from the display image on the display screen.

(Table Processing Method) Next, the table processing method will be described in detail. In this embodiment, the table information of the table configured by the item frame and the data frame in which the entry contents belonging to the item name described in the item frame are described is displayed as the table display information of the image display information. A description will be given of a process performed when displaying the item name and the entry content as the recognition result of the item name and the entry content, respectively, and the obtained character recognition result as the recognition display information of the image display information. FIG. 2 is a flowchart illustrating a table processing flow according to the embodiment.

FIG. 3 is a diagram showing a table to be processed in this embodiment. The table has an item box 134 and a data box 136 as table information. Item frame 13
4, the item name 138 is described. In the data frame 136, the entry content 140 is described. In this embodiment, a horizontal writing table in which the direction of a character line is a horizontal direction (main scanning direction) is to be processed. Note that the display image when this table is read and displayed on the display screen corresponds to the form of FIG. 3.
The display image will be described with reference to the table of FIG.

First, an image of the table shown in FIG. 3 is input by the image input means 10 (S1 in FIG. 2). When inputting an image, the information medium including the table to be processed is optically moved in the main scanning direction (horizontal direction: direction indicated by arrow X in FIG. 3) and the sub-scanning direction (vertical direction: direction indicated by arrow Y in FIG. 3). The optical signal from this information medium is photoelectrically converted by optical scanning.
Here, the X axis in the main scanning direction and the Y axis in the sub scanning direction are virtually set, and the position of the pixel data is represented by the XY coordinate system.

The image image obtained by the photoelectric conversion, that is, the table image information, is stored in the image memory as the table image storage means 12. In the stored image, the position of the pixel is determined by, for example, the control input unit 3 in the XY coordinate system.
The pixel data at the coordinates can be read from the image memory by using 2 or by automatically specifying the coordinates.

Next, ruled lines are extracted from the image by the ruled line extracting means 14 (S2 in FIG. 2). First, a horizontal ruled line, which is a horizontal ruled line, is extracted by a ruled line extracting means 14 for a part or an entire image area stored in a table image storing means 12 by a ruled line extracting means 14. The above continuous black pixels are detected. And
The run length of the detected black pixels is used as a horizontal ruled line, and the coordinate values of the start point and the end point are stored in the table information storage means 24. still,
When extracting a ruled line only in a part of the image image, it is preferable that the area be specified by an operator using the keyboard or the mouse 32, for example. The control input unit 32 can be used to set an arbitrary appropriate value to a predetermined length in the horizontal direction for determining the continuous length of the black pixels as a horizontal ruled line.

In the extraction of the vertical ruled line, which is a vertical ruled line, the ruled line extracting means 14 applies a certain length to the region on the image image in the sub-scanning direction, as in the case of the horizontal ruled line. Detect black pixels that are longer than this. And
Using the length of the detected black pixels as a vertical ruled line, the coordinate values of the start point and the end point are stored in the table information storage means 24. Regarding the constant length in the horizontal direction for determining the continuous length of black pixels as a vertical ruled line, any appropriate value can be set using the control input means 32 individually with the constant length in the vertical direction. .

Next, the table inner frame is extracted by the table inner frame extracting means 16 (S3 in FIG. 2). In extracting a frame in the table (also simply referred to as a frame), first, a command is given by the keyboard or the mouse 32 to read the coordinate values of the start point and the end point of each of the horizontal ruled line and the vertical ruled line from the table information storage means 24. By combining the horizontal ruled line and the vertical ruled line based on the read coordinate values and extracting the coordinates of the intersection, the coordinates of the four intersections (four vertexes of the frame) are defined as one set. Each table frame surrounded by ruled lines is extracted. The extracted coordinates of the four vertices of the inner table frame are stored in the table information storage unit 24. The number of inner frames in the display image corresponding to the table of FIG. 3 is 14.

Next, an item frame and a data frame are extracted from these table frames. Therefore, a keyboard or mouse 32
, An item frame 134 is extracted by the item frame extracting means 18, and a data frame 136 is extracted by the data frame extracting means 20 (S4 in FIG. 2). Here, there are seven item frames 134 and seven data frames 136. FIG.
The flowchart of [1] shows an extraction flow of an item frame and a data frame.

In extracting the item frame 134, first,
The length of the vertical ruled line extending in the vertical direction is compared with a predetermined vertical reference value to extract a ruled line whose length is longer than the predetermined vertical reference value (S6 in FIG. 4). This reference value is
The information may be stored in the table information storage means 24 in advance and read out from the table information, or may be input by operating the keyboard or mouse 32 as required. In the case of the table shown in FIG. 3, all vertical ruled lines correspond to ruled lines having a length equal to or longer than the vertical reference value.

Then, (a) a frame adjacent to the right side of the extracted vertical ruled line is extracted (S7 in FIG. 4). This makes use of the regularity that, in the case of a horizontally written table, the item frame 134 is often located on the right side of a vertical ruled line having a predetermined length or more. Then, the following process (b) is performed on the frame that meets this condition (the process proceeds to S8 in FIG. 4).

(B) A frame whose length (width) in the direction (horizontal direction in this case) along the direction of the character line is within the range of the horizontal reference value is extracted (FIG. 4). S8). The horizontal reference value is also stored in the table information storage means 24 in a freely readable manner in advance, or is input by the mouse 32 or the like as needed.

This utilizes the regularity that the length of the item frame 134 in the row direction is generally shorter than the length of the data frame 136 in the row direction. In this embodiment, a の width of the vertical length of the frame is empirically adopted as the horizontal reference value. Then, the following process (c) is further performed on the frame that satisfies this condition (proceed to S9 in FIG. 4). If the above condition is not met, the process proceeds to step 12 (S12 in FIG. 4) described later.

(C) A frame whose area is within the range of the reference value is extracted (S9 in FIG. 4). This utilizes the regularity that the area of the item box 134 is generally smaller than the area of the data box 136. Then, the processing of the next step 10 is performed on the frame that meets this condition, and if the condition is not met, the process proceeds to step 12.

In this way, a frame that meets the conditions in each of the processes (a), (b) and (c) is extracted as the item frame 134.

Next, character recognition is performed on the item names described in the extracted item frame 134 by using a conventionally known technique (S10 in FIG. 4). The character recognizing unit 22 first stores the item frame 134 stored in the table information storing unit 24.
Based on the coordinates, the image corresponding to this item frame, that is, the image existing in this item frame is obtained from the image image which is the table image information stored in the table image storage means 12, and then, A character pattern is extracted from this image, and finally, a character (item name) is output by comparing the extracted characteristic of the character pattern with a dictionary prepared in advance in the character recognition unit 24. . The character output from the character recognition means 22 is recognition information,
This recognition information is stored in the table information storage means 24 in the item box 134.
Is stored as the recognition result of the item name in association with the coordinate information of the item name. Note that the character pattern may be extracted using a conventionally known technique, for example, based on the continuity of the peripheral distribution values of the image within the frame.

The data frame extracting means 20 extracts a data frame 136 corresponding to the extracted item frame 134,
Character recognition of the entry content described in the data frame 136 is performed (S11 in FIG. 4). In this case, a frame adjacent to the right side of the extracted item frame 134 is automatically extracted as a data frame 136 in association with the item frame. Character recognition in the extracted data frame 136 is performed by the character recognition unit 22 in the same manner as the above-described item name recognition.
Then, the character output by the character recognizing unit 22, that is, the recognition information is associated with the coordinate information of the data frame 136,
It is stored in the table information storage means 24 as a recognition result of the entry contents.

The extraction flow of the item frame and the data frame described above (S7 to S11 in FIG. 4) is performed similarly for all the frames constituting the table. Then, each time the processing is completed for one frame, the control unit 34 sets the in-table frame extraction unit 1
With reference to the total number of frames detected in step 6 and stored in the table information storage means 24, it is determined whether or not processing has been performed for all frames (S12 in FIG. 4). When the processing has been performed for all the frames, the extraction flow of the item frame and the data frame is terminated, and when the processing has not been performed for all the frames, the next target frame is selected (FIG. 4 and S13), and the flow of extracting the item frame and the data frame for the frame again (FIG. 4).
S7 to S11) are performed.

In the present invention, the extracted data frame 13
6 has a special color at the point where the display format is determined and displayed (S5 in FIG. 2). The flowchart of FIG. 5 shows a display format determination flow.

When the recognition results of the item name recognition result and the entry content recognition result are stored in the table information storage means 24, the display control means 30 starts operating in response to the storage.

First, the display control means 30
The image display information to be displayed on the screen is designated, the corresponding image information is read from the table image storage means 12, the display table 26 is searched with the read image information, and the corresponding display format is read. The read display format and the image information are sent to the display means 28, and the image display information is controlled so as to be displayed in the display format described above.

Now, an example of a method of specifying the display format of the data frame 136 as table display information on the screen will be described. In this case, the display control unit 30 reads the coordinate values of one data frame 136 as table image information from the table information storage unit 24 (S14 in FIG. 5). Next, the display control unit 30 reads the recognition result of the item name as another table image information stored in the table information storage unit 24 in association with the read data frame 136 (S15 in FIG. 5). Then, the display control means 30 searches the display format of the data frame stored in the display table 26 based on the recognition result of the read item name (S16 in FIG. 5).

As described above, the display table 26 includes:
For example, if the display color of the data frame is set as the display format, the item name is "red" for "product name", the item name is "blue" for "price", and so on. Information of the corresponding display color is stored. In this way, the display format of the data frame corresponding to the item name is determined. Information on the determined display format is output to the display unit 28.

The display format information stored in the display table 26 is not limited to color information. For example, when the item name is “product name”, the item name is “single underline”, and the item name is “price”. ", A double underline may be displayed below the character image of the data frame, or a dot pattern when the item name is" product name "and an item name" price ". In the case of, a format in which a hatched pattern is superimposed on a data frame and displayed may be used.

Next, the display means 28 is used for the front image storage means 1.
In the display screen displaying the image image (table image information) stored in the table 2 and the character recognition result (recognized image information) stored in the table information storage means 24 as screen display information, as described above. The data frame is displayed in the display format determined as described above (S17 in FIG. 5). For example,
The data frame corresponding to the item name "product name" is colored red, and the data frame corresponding to the item name "price" is colored blue.

The display format determination flow described above (S14 to S17 in FIG. 5) is performed in the same manner as in the above-described example for all data frames constituting the table. Then, every time the processing is completed for one data frame, the control means 34
Referring to the number of data frames detected by the in-table frame extracting means 16 and stored in the table information storage means 24, it is determined whether or not the processing has been performed for all data frames (S18 in FIG. 5).
Then, when the processing has been performed for all data frames,
When the display format determination flow is completed, and the processing has not been performed for all the data frames, the next target data frame is selected (S19 in FIG. 5), and the display format determination flow for that data frame is again performed (S19 in FIG. 5). Steps S14 to S17 in FIG. 5 are performed.

FIG. 6 is a diagram showing a display image of the table to be processed shown in FIG. 3 on the display screen and a recognition display image of the character recognition result. The upper part in the drawing of FIG. 6 shows a display image 142 displayed based on the image image (table image information) stored in the table image storage unit 12. Therefore, the display image 142 shows the above-described display format, that is, the data frame 136 in a display format corresponding to the item name 138 to which the entry 140 described in the data frame 136 belongs. In the lower part of the drawing of FIG. 6, the item name 138 and the entry content 14 described in the item box 134 and the data box 136, respectively, are shown.
Recognition display information that is the character recognition result 144 of 0, that is, the recognition result 146 of the item name and the recognition result 148 of the entry content
Is shown. In this embodiment, each data frame 136 on the image is displayed in a different mask pattern display format. For example, the data frame whose entry is "〇〇 company" is displayed in a display format determined or assigned by the item name "manufacturer" to which the entry belongs, that is, in this case, in a hatched pattern inclined from upper left to lower right. ing. Similarly, the data frame whose entry is "ABCDEF" is displayed in a diagonal pattern inclined from upper right to lower left assigned by the item name "product name". In addition, for example, the data frame whose entry content is “$ 1,234” is similarly displayed by a diagonal pattern inclined from upper left to lower right. A data frame having no entry is blank.

As described above, in this example, since the data frame is displayed in a display format depending on the item name to which the entry belongs, the display image 142 is compared with the recognition display image 144 of the character recognition result. When confirming the correctness of the character recognition result 144 and subsequently correcting the result, it is easy to take correspondence between them. Specifically, the operator recognizes the recognition result 144.
Focusing on the data frame 13 corresponding to the recognition result 148 of the entry content at an unspecified position in the table area of the display image 142
In the case of searching for the data pattern 13, the mask pattern to be displayed is determined by the item name 138 regardless of the format of the displayed table.
6 can be specified. Therefore, the visibility of the operator is improved, and operation errors are reduced.

[Second Embodiment] (Table Processing Apparatus) A table processing apparatus according to a second embodiment
Only the data information stored in the display table 26 is different from the configuration (FIG. 1) of the table processing apparatus described in the embodiment. Only the behavior is different. Therefore,
Unless necessary, duplicate description will be omitted.

In the display table 26, the display format of the item frame is set so as to correspond to the entry contents belonging to the item name described in this item frame and to be different depending on the individual entry content. For example, information such as displaying an item frame in which the item name to which the entry “B” belongs is displayed in a display format of “b” is stored, and a recognition result of the entry of the entry “B” is input. And that information is output.

The display control means 30 displays the display format of the item frame corresponding to the recognition result of the entry stored in the table information storage means 24 when displaying the image as a display image on the display screen. This is a means for searching the table 26 and instructing the display means 28. For example,
The recognition result of the entry "B" is read out from the table information storage means 24, and the display format "b" of the item frame corresponding to the entry "B" is extracted from the display table 26 and instructed to the display means 28. I do. Then, the display unit 28 displays the item frame corresponding to the entry “B” in the display format “b”.

As described above, since the display format of the item frame on the image image is displayed according to the corresponding entry contents, it is easy to find the item name corresponding to the recognition result of the item name of the character recognition result from the image image. .

(Table Processing Method) The table processing method according to the second embodiment differs from the table processing method described in the first embodiment only in the display format determination flow. I do.

FIG. 7 is a flowchart showing a display format determination flow according to the second embodiment. With reference to this flow, the step (S5 in FIG. 2) of determining and displaying the display format of the extracted item frame 134 according to the item frame and data frame extraction flow (FIG. 4) will be described.

First, when the display control means 30 operates, the display control means 30 reads the coordinate value of one item frame 134 from the table information storage means 24 (S20 in FIG. 7). Next, the display control unit 30 reads the recognition result of the entry content stored in the table information storage unit 24 in association with the read item frame 134 (S21 in FIG. 7). Then, the display control means 30 searches the display format of the item frame stored in the display table 26 based on the recognition result of the read entry contents (S22 in FIG. 7).

As described above, the display table 26 includes:
For example, if the display color of the item frame is set as the display format, the entry content is “red” for “ABCDEF”,
The information of the display color corresponding to the entry content is stored such that the entry content is “blue” for “$ 1,234”. In this way, the display format of the item frame corresponding to the entry content is determined by the entry content corresponding to the item frame. Information on the determined display format is output to the display unit 28.

Next, the display means 28 is used for the front image storage means 1.
2 and the table information storage means 2
The item frame is displayed in the determined display format on the display screen displaying the character recognition result stored in No. 4 (S23 in FIG. 7). For example, if the entry is "AB
The item frame corresponding to “CDEF” is surrounded by a red frame, and the item frame corresponding to “$ 1,234” to be entered is surrounded by a blue frame.

The display format determination flow described above (S20 to S23 in FIG. 7) is performed in the same manner for all the item frames constituting the table. Each time the processing for one item frame is completed, the control unit 34 refers to the number of item frames detected by the in-table frame extracting unit 16 and stored in the table information storage unit 24, and performs processing for all the item frames. Is determined (S24 in FIG. 7). Then, if the processing has been performed for all the item frames, the display format determination flow ends, and if the processing has not been performed for all the item frames yet, the next processing target item frame is selected (FIG. 7 S25),
Again, the display format determination flow (S20 to S23 in FIG. 7) is performed for the item frame.

FIG. 8 is a diagram showing a display image on the display screen and a recognition display image of the character recognition result of the table to be processed shown in FIG. The display image 14 stored in the table image storage unit 12 is shown in the upper part of FIG.
2 is displayed in the display format described above, that is, the item frame 134 is displayed in a display format corresponding to the entry content 140 belonging to the item name 138 described in the item frame 134. In the part, the item name 138 described in each of the item frame 134 and the data frame 136
And the character recognition result 144 of the entry 140, ie, the item name recognition result 146, and the entry content recognition result 148.
Is shown. In this embodiment, a different mask pattern is displayed on each item frame 134 on the display image in a display format. Therefore, when the display image 142 and the character recognition result (recognition display image) 144 are compared with each other to check whether the character recognition result 144 is correct or not and to correct the result thereafter,
The correspondence between these is easy to take. Specifically, when the operator looks at the character recognition result 144 and searches for the item frame 134 corresponding to the item name recognition result 146 at an unspecified position in the table area of the display image 142, the displayed table is displayed. Regardless of the format, since the mask pattern to be displayed is determined by the entry 140, the operator can easily specify the position of the item frame 134. Therefore, the visibility of the operator is improved, and operation errors are reduced.

[Third Embodiment] (Table Processing Apparatus) The table processing apparatus according to the third embodiment
Only the data information stored in the display table 26 is different from the configuration (FIG. 1) of the table processing apparatus described in the embodiment. Only the behavior is different. Therefore,
Unless necessary, duplicate description will be omitted.

In the display table 26, the display format of the recognition result of the entry content is made to correspond to the recognition result of the item name associated with the recognition result of the entry content. Has been set. For example, information such as displaying the recognition result of the entry content associated with the recognition result “A” of the item name in the display format “a” is stored, and the recognition result “A” of the item name is input. And that information is output.

The display format includes, for example, a format in which a mask pattern is superimposed on the recognition result of the input content, and a format in which characters of the recognition result of the input content are colored. Further, a format in which the font of the character is changed for each recognition result of the entry content may be used. For example, in the case of a display format in which the recognition result of the entry is surrounded by a colored frame, the color of the frame surrounding the recognition result of each entry is set to a different color for each recognition result of the entry.

When displaying the image, the display control means 30 searches the display table 26 for a display format of the recognition result of the entry corresponding to the recognition result of the item name stored in the table information storage means 24. And instructs the display means 28. For example, the recognition result “A” of the item name is read from the table information storage unit 24, and the display format “a” of the recognition result of the entry corresponding to the recognition result “A” of the item name is displayed.
From the display table 26 and instruct the display means 28. Then, the display means 28 displays the recognition result of the entry content corresponding to the recognition result "A" of the item name in the display format "a".

As described above, the display format of the recognition result of the entry content as the character recognition result is displayed according to the recognition result of the corresponding item name, so that the recognition result of the entry content corresponding to the entry content of the display image is displayed. , Easy to find in the displayed image.

(Table Processing Method) The table processing method according to the third embodiment differs from the table processing method described in the first embodiment only in the display format determination flow. I do.

FIG. 9 is a flowchart showing a display format determination flow according to the third embodiment. With reference to this flow, the step (S5 in FIG. 2) of determining and displaying the display format of the recognition result of the entry content extracted according to the extraction flow of the item frame and the data frame (FIG. 4) will be described.

First, the display control means 30 reads the recognition result of a set of item names and the recognition result of entry contents from the table information storage means 24 (S27 in FIG. 9). Then, the display control means 30 determines, based on the recognition result of the read item name,
The display format of the recognition result of the entry stored in the display table 26 is searched (S28 in FIG. 9).

As described above, in the display table 26,
For example, if the text color of the recognition result of the entry is set as the display format, the recognition result of the item name is "red" for "product name", and the recognition result of the item name is "red" for "price". For example, "blue" stores information on the character display color corresponding to the recognition result of the item name. In this way, the display format of the recognition result of the entry content corresponding to the recognition result of the item name is determined. Information on the determined display format is output to the display unit 28.

Next, the display means 28 is used for the front image storage means 1.
2 and the table information storage means 2
On the display screen displaying the character recognition result stored in the display unit 4 as a display image and a recognition display image, the recognition result of the entered content is displayed in the determined display format (S29 in FIG. 9). ). For example, the recognition result of the entry corresponding to the item name "product name" is surrounded by a red frame, and the recognition result of the entry corresponding to the item name is "price" is surrounded by a blue frame. .

The display format determination flow described above (S27 to S29 in FIG. 9) is similarly performed for the recognition results of all the entries constituting the table. Each time the processing for the recognition result of one entry is completed, the control means 3
4. Table information storage means 24 detected by table inner frame extraction means 16
It is determined whether or not the processing has been performed on the recognition results of all the input contents with reference to the number of recognition results of the input contents stored in (2) (S30 in FIG. 9). Then, if the processing has been performed on the recognition results of all the entries, the display format determination flow ends. If the processing has not been performed on the recognition results of all the entries yet, the next entry Is selected (S31 in FIG. 9), and the display format determination flow (S27 to S2 in FIG. 9) is again performed for the recognition result of the entry.
Perform 9).

FIG. 10 is a diagram showing a display image and a recognition display image of the processing target table shown in FIG. 3 on the display screen, that is, a character recognition result. In the upper part of FIG. 10, the display image 14 stored in the front image storage unit 12 is displayed.
2 is the display format described above, that is, the recognition result 1 of the entry contents.
48 is displayed in a display format corresponding to the recognition result 146 of the item name associated with the recognition result 148 of the entry, and the lower part of FIG. The character recognition result 144 of the item name 138 and the entry content 140 described in each of the data frames 136, that is, the item name recognition result 146 and the entry content recognition result 148 are shown. In this embodiment,
Different mask patterns are superimposed on the recognition result 148 of each entry on the character recognition result.
Therefore, when the display image 142 is collated with the character recognition result 144 to confirm the correctness of the character recognition result 144 and to subsequently correct the result, it is easy to take a correspondence between them. Specifically, when the operator pays attention to the display image 142 and searches for the recognition result 148 of the entry corresponding to the data frame 136 and the entry 140 at an unspecified position in the table area of the display image 142, Regardless of the format of the displayed table, since the recognition result 146 of the item name, that is, the mask pattern to be displayed is determined by the item name, the operator can easily specify the position of the recognition result 148 of the entry. Therefore, the visibility of the operator is improved, and operation errors are reduced.

[Fourth Embodiment] (Table Processing Apparatus) The table processing apparatus according to the third embodiment
Only the data information stored in the display table 26 is different from the configuration (FIG. 1) of the table processing apparatus described in the embodiment. Only the behavior is different. Therefore,
Unless necessary, duplicate description will be omitted.

In the display table 26, the display format of the recognition result of the item name is made to correspond to the recognition result of the entry content associated with the recognition result of the item name. Has been set. For example, information such that the recognition result of the item name associated with the recognition result “B” of the entry is displayed in the display format “b” is stored, and the recognition result “B” of the entry is input. And that information is output.

The display control means 30 searches the display table 26 for the display format of the recognition result of the item name corresponding to the recognition result of the entry stored in the table information storage means 24 when displaying the image. And instructs the display means 28. For example, the recognition result "B" of the entry content
Is read out from the table information storage means 24, and the display format "b" of the recognition result of the item name corresponding to the recognition result "B" of the entry content is extracted from the display table 26 and displayed on the display means 28.
To instruct. Then, the display unit 28 displays the recognition result of the item name corresponding to the recognition result “B” of the entry content in the display format “b”.

As described above, the display format of the recognition result of the item name on the character recognition result is displayed according to the recognition result of the corresponding entry content. Easy to find from recognition results.

(Table Processing Method) The table processing method according to the fourth embodiment differs from the table processing method described in the first embodiment only in the display format determination flow. I do.

FIG. 11 is a flowchart showing a display format determination flow according to the fourth embodiment. With reference to this flow, the step (S5 in FIG. 2) of determining and displaying the display format of the recognition result of the extracted item name according to the extraction flow of the item frame and the data frame (FIG. 4) will be described.

First, the display control means 30 reads a recognition result of a set of item names and a recognition result of entry contents from the table information storage means 24 (S32 in FIG. 11). Then, the display control unit 30 searches the display format of the recognition result of the item name stored in the display table 26 based on the read result of the entry content (S33 in FIG. 11).

As described above, in the display table 26,
For example, when the character color of the recognition result of the item name is set as the display format, the recognition result of the entry is "ABCDE".
"F" for "red" and the recognition result of the entry is "$ 1,
For example, "blue" for "234" is stored as character display color information corresponding to the recognition result of the input content.
In this way, the display format of the recognition result of the item name corresponding to the recognition result of the entry content is determined. Information on the determined display format is output to the display unit 28.

Next, the display means 28 is used for the front image storage means 1.
In the display screen displaying the display image stored in the display information 2 and the character recognition result stored in the table information storage means 24, the recognition result of the item name is displayed in the determined display format. (S34 of FIG. 11). For example, the recognition result of the item name whose recognition result of the entry corresponds to “ABCDEF” is surrounded by a red frame, and the recognition result of the entry is “$ 1,
The recognition result of the item name corresponding to “234” is surrounded by a blue frame.

The display format determination flow described above (FIG. 11)
S32 to S34) are similarly performed on the recognition results of all the item names constituting the table. Each time the processing on the recognition result of one item name is completed, the control unit 34
Is detected by the table inner frame extracting means 16 and the table information storing means 24
It is determined whether or not the processing has been performed on the recognition results of all the item names with reference to the recognition results of the item names stored in (1) (S35 in FIG. 11). If the processing has been performed on the recognition results of all the item names, the display format determination flow ends, and if the processing has not been performed on the recognition results of all the item names, the next target item name has not been processed. Is selected (S36 in FIG. 11), and the display format determination flow (S32 to S34 in FIG. 11) is performed again on the recognition result of the item name.

FIG. 12 is a diagram showing a display image of the table to be processed shown in FIG. 3 and a character recognition result. In the upper part of FIG. 12, the display image 142 stored in the table image storage unit 12 is associated with the above-described display format, that is, the item name recognition result 146 is associated with the item name recognition result 146. FIG. 12 shows a display in a display format corresponding to the recognition result 148 of the entry content. In the lower part of FIG. 12, the item names 138 and 136 described in the item box 134 and the data box 136 are shown. A character recognition result 144 of the entry content 140, that is, an item name recognition result 146 and an entry content recognition result 148 are shown. In this embodiment, the recognition result 14 of each item name on the character recognition result
6 is displayed in a display format in which different mask patterns are overlapped. Therefore, the display image 142 and the character recognition result 14
4 is easily checked when the character recognition result 144 is checked for correctness by comparing the character recognition result 144 with the subsequent result correction. Specifically, when the operator looks at the display image 142 and searches for the recognition result 146 of the item frame 134 and the item name corresponding to the item name 138 at an unspecified position in the table area of the display image 142, Regardless of the format of the displayed table, since the recognition result 148 of the entry content, that is, the mask pattern to be displayed is determined by the entry content, the operator can easily recognize the recognition result 14 of the item name.
6 can be specified. Therefore, the visibility of the operator is improved, and operation errors are reduced.

[Fifth Embodiment] (Table Processing Apparatus) The table processing apparatus according to the fifth embodiment
Only the data information stored in the display table 26 is different from the configuration (FIG. 1) of the table processing apparatus described in the embodiment. Only the behavior is different. Therefore,
Unless necessary, duplicate description will be omitted.

In the display table 26, the display format of the data frame and the recognition result of the entry content are made to correspond to the recognition result of the item name associated with the recognition result of the entry content, and the recognition result of the individual item name is displayed. They are set differently depending on the situation.
For example, it stores information such as displaying the data frame associated with the item name recognition result “A” and the recognition result of the entry content in the display format “a”, and the item name recognition result “A” is stored. When the information is input, the information is output.

When displaying the image, the display control means 30 displays the data frame corresponding to the recognition result of the item name stored in the table information storage means 24 and the display format of the recognition result of the entry contents in the display table. It is a means for searching from 26 and instructing the display means 28. For example, the recognition result “A” of the item name is read from the table information storage unit 24, and the display format “a” of the recognition result of the data frame and the entry content corresponding to the recognition result “A” of the item name is displayed in the display table 26. And instruct the display means 28. Then, the display unit 28 displays the data frame corresponding to the recognition result “A” of the item name and the recognition result of the entry in the display format “a”.

As described above, the display format of the data frame on the image image and the display format of the recognition result of the entry on the character recognition result are displayed according to the recognition result of the corresponding item name. It is easy to find the recognition result of the entry corresponding to the content from the character recognition result.

(Table Processing Method) The table processing method according to the fifth embodiment differs from the table processing method described in the first embodiment only in the display format determination flow. I do.

FIG. 13 is a flowchart showing a display format determination flow according to the fifth embodiment. With reference to this flow, the step (S5 in FIG. 2) of determining and displaying the display format of the recognition result of the data frame and the entry content extracted according to the extraction flow of the item frame and the data frame (FIG. 4) will be described.

First, the display control means 30 reads the coordinate value of one data frame from the table information storage means 24 (FIG. 13).
S37). Next, the display control unit 30 reads the recognition result of the item name and the recognition result of the entry content stored in the table information storage unit 24 in association with the read data frame (S38 in FIG. 13). And display control means 3
0 searches the display format of the recognition result of the data frame and the entry content stored in the display table 26 based on the recognition result of the read item name (S39 in FIG. 13).

As described above, the display table 26 contains
For example, if the color in the frame of the data frame and the character color of the recognition result of the entry are set as the display format, the recognition result of the item name is “red” for “product name” and the recognition result of the item name is “red”. Information such as the in-frame display color and the character display color corresponding to the recognition result of the item name is stored, such as “blue” for “price”. In this way, the data frame corresponding to the item name recognition result and the display format of the entry content recognition result are determined. Information on the determined display format is output to the display unit 28.

Next, the display means 28 is the front image storage means 1.
In the display screen displaying the display image stored in the table 2 and the character recognition result stored in the table information storage means 24, the recognition result of the data frame and the entry content is displayed in the determined display format. It is displayed (S40 in FIG. 13).
For example, a data frame whose item name recognition result corresponds to "product name" and the recognition result of the entry content are surrounded by a red frame, and a recognition result of the item name corresponds to "price" and the data frame and the recognition result of the entry content. Is surrounded by a blue frame.

The display format determination flow described above (FIG. 13)
S37 to S40) are performed in the same way for the recognition results of all data frames and entry contents that make up the table. Then, each time the processing of one set of data frame and the recognition result of the entry content is completed, the control unit 34 causes the table frame extraction unit 1
With reference to the number of data frames detected in step 6 and stored in the table information storage means 24, it is determined whether or not processing has been performed for all data frames (S41 in FIG. 13). If the processing has been performed on all the data frames, the display format determination flow ends, and if the processing has not been performed on all the data frames, the next target data frame is selected (see FIG. 13). S4
2) Again, a display format decision flow (S in FIG.
37 to S40).

FIG. 14 is a diagram showing a display image and a character recognition result of the table to be processed shown in FIG. In the upper part of FIG. 14, the display image 142 stored in the table image storage unit 12 is displayed in the display format described above, that is, the data frame 136 and the recognition result 1 of the entry content of the data frame 136.
14 is displayed in a display format corresponding to the recognition result 146 of the item name associated with the recognition result 148 of the entry content, and the lower part in FIG. In addition, the character recognition result 144 of the item name 138 and the entry content 140 described in each of the data frames 136, that is, the item name recognition result 146 and the entry content recognition result 148 are shown. In this embodiment, a different mask pattern is displayed on each data frame 136 on the display image and the recognition result 148 of each entry on the character recognition result. Therefore, when the display image 142 is collated with the character recognition result 144 to confirm the correctness of the character recognition result 144 and to subsequently correct the result, it is easy to take a correspondence between them. Specifically, the operator pays attention to the display image 142, and
2 is a recognition result 148 of the entry corresponding to the data frame 136 and the entry 140 at an unspecified position in the table area.
In the case of searching for a name, the operator easily specifies the position of the recognition result 148 of the entry contents because the recognition result 146 of the item name, that is, the mask pattern to be displayed is determined by the item name regardless of the format of the displayed table. be able to. Further, if the display format of the data frame and the recognition result of the entry content corresponding to the recognition result of a certain item name are associated with each other, the collation between the image image and the character recognition result is further facilitated. Therefore, the visibility of the operator is improved, and operation errors are reduced.

[Sixth Embodiment] (Table Processing Apparatus) The table processing apparatus according to the sixth embodiment
Only the data information stored in the display table 26 is different from the configuration (FIG. 1) of the table processing apparatus described in the embodiment. Only the behavior is different. Therefore,
Unless necessary, duplicate description will be omitted.

In the display table 26, the display format of the recognition result of the item frame and the item name is made to correspond to the recognition result of the entry content associated with the recognition result of the item name, and the recognition result of the individual entry content is displayed. They are set differently depending on the situation. For example, information such that the recognition result of the item frame and the item name associated with the recognition result “B” of the entry content is displayed in a display format of “b” is stored, and the recognition result “B” of the entry content is stored. When the information is input, the information is output.

The display control means 30 displays the display format of the item frame and the item name recognition result corresponding to the recognition result of the entry content stored in the table information storage means 24 when displaying the image. It is a means for searching from 26 and instructing the display means 28. For example, the recognition result "B" of the entry content is read from the table information storage means 24, and the display format "b" of the recognition result of the item frame and the item name corresponding to the recognition result "B" of the entry content is displayed in the display table 26. And instruct the display means 28. And display means 2
8 displays the recognition result of the item frame and the item name corresponding to the recognition result “B” of the entry content in the display format of “b”.

As described above, the display format of the item frame on the image image and the display format of the recognition result of the item name on the character recognition result are displayed in accordance with the recognition result of the corresponding entry contents. It is easy to find the recognition result of the item name corresponding to the name from the character recognition result.

(Table Processing Method) The table processing method according to the sixth embodiment differs from the table processing method described in the first embodiment only in the display format determination flow. I do.

FIG. 15 is a flowchart showing a display format determination flow according to the sixth embodiment. With reference to this flow, a step (S5 in FIG. 2) of determining and displaying the display format of the recognition result of the item frame and item name extracted according to the item frame and data frame extraction flow (FIG. 4) will be described.

First, the display control means 30 reads the coordinate value of one item frame from the table information storage means 24 (S43 in FIG. 15). Next, the display control unit 30 reads the recognition result of the item name and the recognition result of the entry content stored in the table information storage unit 24 in association with the read item frame (S44 in FIG. 15). Then, the display control means 30
Retrieves the display format of the recognition result of the item frame and the item name stored in the display table 26 based on the recognition result of the read entry (S45 in FIG. 15).

As described above, in the display table 26,
For example, when the color in the frame of the item box and the character color of the recognition result of the item name are set as the display format, the recognition result of the entry content is “red” for “ABCDEF” and the recognition result of the entry content is “red”. For example, “blue” for “¥ 1,234” stores information on the in-frame display color and the character display color corresponding to the recognition result of the entry content. In this way,
The display format of the item frame and the item name recognition result corresponding to the recognition result of the entry content is determined. Information on the determined display format is output to the display unit 28.

Next, the display means 28 is the front image storage means 1.
In the display screen displaying the display image stored in the display information storage unit 2 and the character recognition result stored in the table information storage means 24, the recognition result of the item frame and the item name is displayed in the determined display format. It is displayed (S46 in FIG. 15). For example, the recognition result of the entry frame and the item name whose recognition result corresponds to “ABCDEF” is surrounded by a red frame, and the recognition result of the entry content corresponds to “$ 1,234”. The recognition result is surrounded by a blue frame.

The display format decision flow described above (FIG. 15)
S43 to S46) are similarly performed on the recognition results of all the item frames and item names constituting the table. And
Each time the processing of the recognition result of a set of item frames and item names is completed, the control unit 34 refers to the number of item frames detected by the in-table frame extraction unit 16 and stored in the table information storage unit 24, It is determined whether or not processing has been performed for all item frames (S47 in FIG. 15). Then, when the processing has been performed for all item frames, the display format determination flow ends,
If the processing has not been performed for all the item frames, the next target item frame is selected (S48 in FIG. 15), and the display format determination flow is again performed for the recognition result of the item frame and the corresponding item name. (S43 to S47 in FIG. 15) are performed.

FIG. 16 is a diagram showing an image image and a character recognition result of the table to be processed shown in FIG. FIG.
In the upper part of the figure, the display image 142 stored in the table image storage means 12 is displayed in the display format, that is, the item frame 134 and the recognition result 14 of the item name of the item frame 134.
6 is displayed in a display format corresponding to the recognition result 148 of the entry content associated with the recognition result 146 of the item name. The lower part in FIG. In addition, the character recognition result 144 of the item name 138 and the entry content 140 described in each of the data frames 136, that is, the item name recognition result 146 and the entry content recognition result 148 are shown. In this embodiment,
Different mask patterns are displayed on the respective item frames 134 on the image and the recognition results 146 of the respective item names on the character recognition result in a display format in which different mask patterns are overlapped. Therefore, when the image image 142 is compared with the character recognition result 144 to confirm the correctness of the character recognition result 144 and to subsequently correct the result, it is easy to take a correspondence between them. Specifically, the operator pays attention to the display image 142, and
When the search result 146 of the item name corresponding to the item frame 134 and the item name 138 at an unspecified position in the table area of No. 2 is found, regardless of the format of the displayed table,
Since the recognition result 148 of the entry content, that is, the mask pattern to be displayed is determined by the entry content, the operator can easily specify the position of the item name recognition result 146. Further, if the display format of the item frame and the item name recognition result corresponding to the recognition result of a certain entry content is associated, the collation between the image image and the character recognition result is further facilitated. Therefore, the visibility of the operator is improved,
Operation errors are reduced.

[0141]

According to the table processing method of the first aspect of the present invention, a certain screen display information is displayed in a display format specified by another screen display information. By using the reference, the display position of the above-mentioned certain screen display information on the display screen can be easily understood.

Further, for example, the certain screen display information is used as table display information of a data frame, and the other screen display information is used as recognition display information of an item name belonging to the entry content described in the data frame. Thus, since each data frame is displayed in a display format corresponding to the item name, the position of the entry corresponding to the recognition result of the entry can be easily understood. Therefore, according to the table processing method of the present invention, it is easier to determine whether the character recognition processing result of the entry is correct or not, as compared with the related art.

Further, for example, the certain screen display information is used as table display information of an item frame, and the another screen display information is used as recognition display information of entry contents belonging to an item name described in this item frame. Accordingly, since each item frame is displayed in a display format corresponding to the entry contents, the position of the item name corresponding to the recognition result of the item name can be easily understood. Therefore, according to the table processing method of the present invention, it is easier to determine whether or not the result of the character recognition processing of the item name is correct as compared with the related art.

Further, for example, by using the certain screen display information as the recognition display information of the entry content and the another screen display information as the recognition display information of the item name belonging to the entry content, the item name can be recognized. Since the recognition result of each entry is displayed in a display format corresponding to the result, the position of the recognition result of the entry corresponding to the entry can be easily understood. Therefore, according to the table processing method of the present invention, it is easier to determine whether the character recognition processing result of the entry is correct or not, as compared with the related art.

For example, by using the certain screen display information as the recognition display information of the item name and the another screen display information as the recognition display information of the entry content belonging to this item name, the recognition of the entry content can be performed. Since the recognition result of each item name is displayed in a display format corresponding to the result, the position of the recognition result of the item name corresponding to the item name can be easily understood. Therefore,
According to the table processing method of the present invention, it is easier to determine whether the character recognition processing result of the item name is correct or not, as compared with the related art.

Also, for example, the certain screen display information is used as table display information of a data frame and recognition display information of the entry content described in the data frame, and the another screen display information is used as an item belonging to the entry content. By using the name recognition display information, the recognition results of each data frame and entry contents are displayed in a display format corresponding to the item name recognition result.
The position of the entry corresponding to the recognition result of the entry can be easily understood. Therefore, according to the table processing method of the present invention, it is easier to determine whether the character recognition result of the entered content is correct or not, as compared with the related art.

Also, for example, the certain screen information is used as table display information of an item frame and recognition display information of an item name described in this item frame, and the another screen display information is used as entry contents belonging to this item name. By using the recognition display information of
Since the recognition result of each item frame and item name is displayed in a display format corresponding to the recognition result of the entry content, the position of the item name corresponding to the recognition result of the item name can be easily understood. Therefore, according to the table processing method of the present invention, it is easier to determine whether or not the result of the character recognition processing of the item name is correct as compared with the related art.

Next, according to the table processing apparatus of the second invention according to this application, by specifying the image information, a certain screen display information is stored in the table information storage means. The image information is read out, and the corresponding display format stored in the display table memory is read out and displayed with the image information. Therefore, since a certain screen display information is displayed in a display format specified by another image information, the screen display information of the other image information is used as a position reference, so that a certain screen display in the display screen is displayed. The display position of information is easy to understand.

Further, for example, the certain screen display information is used as table display information of a data frame, and the another image information stored in the display table memory is replaced with a table image of an item name belonging to the entry content described in the data frame. As information, each data frame is displayed in a display format corresponding to the item name. Therefore, if the recognition display information of the item name is used as the position reference, the data frame displayed in the corresponding display format can be easily found. Therefore, according to the table processing apparatus of the present invention, it is easy to determine whether the character recognition processing result of the entry content described in the data frame is correct or not.

Further, for example, the certain screen display information is used as table display information of an item frame, and the another image information stored in the display table memory is replaced with a table image of the entry contents belonging to the item name described in the item frame. If it is information, each item frame is displayed in a display format corresponding to the entry contents. Therefore, if the recognition display information of the entry is used as the position reference, the item frame displayed in the corresponding display format can be easily found. Therefore, according to the table processing device of the present invention, it is easy to determine whether the character recognition processing result of the item name described in the item box is correct or not.

Further, for example, if the certain screen display information is used as recognition display information of entry contents, and the other image information stored in the display table memory is table image information of an item name belonging to the entry contents, The recognition display information of each entry is displayed in a display format corresponding to the name. Therefore, if the recognition display information of the item name is used as the position reference, the entry content displayed in the corresponding display format can be easily found. Therefore, according to the table processing apparatus of the present invention, it is easy to determine whether the character recognition processing result of the entry is correct or not.

For example, if the certain screen display information is used as item name recognition display information, and the another image information stored in the display table memory is used as table image information of the entry contents belonging to the item name, The recognition display information of each item name is displayed in a display format corresponding to the content. Therefore, if the recognition display information of the entry is used as a position reference, the item name displayed in the corresponding display format can be easily found. Therefore, according to the table processing apparatus of the present invention, it is easy to determine whether the character recognition processing result of the item name is correct or not.

Also, for example, the certain screen display information is used as table display information of a data frame and recognition display information of the entry content described in the data frame, and the other image information stored in the display table memory is stored in the display table memory. If the table image information is the item name belonging to the content, the screen display information of the data frame and the entry content is displayed in a display format corresponding to the item name. Therefore, if the recognition display information of the item name is used as the position reference, the data frame and the entry content displayed in the corresponding display format can be easily found. Therefore, according to the table processing apparatus of the present invention, it is easy to determine whether the character recognition processing result of the entry is correct or not.

For example, the certain screen display information is used as table display information of an item frame and recognition display information of an item name described in the item frame, and the other image information stored in a display table memory is stored in the display table memory. Assuming that the tabular image information of the entry content belonging to the name, the screen display information of the item frame and the item name is displayed in a display format corresponding to the entry content. Therefore, if the recognition display information of the entry is used as the position reference, the item frame and the item name displayed in the corresponding display format can be easily found. Therefore, according to the table processing apparatus of the present invention, it is easy to determine whether the character recognition processing result of the item name is correct or not.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating a configuration of a table processing apparatus according to an embodiment.

FIG. 2 is a diagram illustrating a table processing flow according to the embodiment;

FIG. 3 is a diagram illustrating a configuration of a table to be processed;

FIG. 4 is a diagram illustrating an extraction flow of an item frame and a data frame according to the embodiment.

FIG. 5 is a diagram illustrating a display format determination flow according to the first embodiment.

FIG. 6 is a diagram illustrating a display image and a character recognition result according to the first embodiment.

FIG. 7 is a diagram illustrating a display format determination flow according to the second embodiment.

FIG. 8 is a diagram illustrating a display image and a character recognition result according to the second embodiment.

FIG. 9 is a diagram illustrating a display format determination flow according to the third embodiment.

FIG. 10 is a diagram illustrating a display image and a character recognition result according to the third embodiment.

FIG. 11 is a diagram illustrating a display format determination flow according to a fourth embodiment.

FIG. 12 is a diagram illustrating a display image and a character recognition result according to the fourth embodiment.

FIG. 13 is a diagram illustrating a display format determination flow according to the fifth embodiment.

FIG. 14 is a diagram illustrating a display image and a character recognition result according to the fifth embodiment.

FIG. 15 is a diagram illustrating a display format determination flow according to the sixth embodiment.

FIG. 16 is a diagram illustrating a display image and a character recognition result according to the sixth embodiment.

[Explanation of symbols]

 10: Image input means 12: Table image storage means 14: Ruled line extraction means 16: Table frame extraction means 18: Item frame extraction means 20: Data frame extraction means 22: Character recognition means 24: Table information storage means 26: Display table 28: display means 30: display control means 32: control input means 34: control means 36: extraction means 134: item frame 136: data frame 138: item name 140: entry content 142: display image 144: character recognition result 146: item Name recognition result 148: Recognition result of entry

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁶ Identification code FIG06F 15/22 310B

Claims (22)

[Claims] 1. An item frame, an item name described in the item frame,
The table contents of the entry contents belonging to the item name and the data frame in which the entry contents are described are converted into table image information by image input means, and the table image information corresponding to the item names and the entry contents is converted. A character recognition process is performed to obtain an item name recognition result and an entry content recognition result, and the table image information and the recognition result are converted into table display information and screen display information of recognition display information, respectively, and displayed on a screen of a display unit. A table processing method comprising: displaying certain screen display information in a display format specified by another screen display information individually related thereto; 2. The table processing method according to claim 1, wherein the image information is stored in the table information storage means in the certain screen display information in a correspondence relationship in which image information related to the another screen display information can be read. The display format is stored in a display table memory in correspondence with the image information so that the display format specified by the image information can be read out, and the certain screen display information is stored in the screen of the display unit. When the image information of the certain screen display information is read, the image information of the another screen display information is read from the table information storage means with the image information of the certain screen display information, and the display table memory is searched with the read image information to respond. A table processing method, comprising: reading a display format; and providing the read display format to the display unit. 3. The table processing method according to claim 2, wherein a plurality of ruled lines are extracted from the table image information obtained by the conversion by the image input means, and a plurality of tables surrounded by the extracted ruled lines are extracted. Extracting an inner frame, extracting an item frame of the table image information from the extracted table inner frame, extracting the table inner frame adjacent to the extracted item frame as a data frame of the table image information, The item name described in the extracted item frame and the entry content described in the extracted data frame are targeted for the character recognition processing, and the image information stored in the table information storage unit is extracted. A table processing method using the item frame, the extracted data frame, the item name recognition result, and the entry content recognition result. 4. The table processing method according to claim 1, wherein said certain screen display information is table display information of said data frame, and said another screen display information is said table display information. A table processing method, wherein recognition display information of an item name belonging to an entry content described in a data frame is used. 5. The table processing method according to claim 1, wherein the certain screen display information is table display information of the item frame,
And a table processing method, wherein the another screen display information is recognition display information of an entry belonging to an item name described in the item frame. 6. The table processing method according to claim 1, wherein said certain screen display information is used as recognition display information of said entry, and said another screen display information is used as said another screen display information. A table processing method, wherein recognition display information of an item name belonging to an entry is used. 7. The table processing method according to claim 1, wherein the certain screen display information is the item name recognition display information, and the another screen display information is A table processing method, wherein recognition and display information of entry contents belonging to an item name is used. 8. The table processing method according to claim 1, wherein the certain screen display information is recognized as table display information of the data frame and entry contents described in the data frame. A table processing method as display information, and wherein the another screen display information is recognition display information of an item name belonging to the entry. 9. The table processing method according to claim 1, wherein the certain screen display information is recognized by the table display information of the item frame and the item name described in the item frame. A table processing method as display information, and wherein the another screen display information is recognition display information of an entry belonging to the item name. 10. An image input means for converting each table information of an item frame, an item name described in the item frame, an entry content belonging to the item name, and a data frame in which the entry content is described into tabular image information. Character recognition means for performing character recognition on the table image information corresponding to the item name and the entry content and acquiring the recognition result of the item name and the recognition result of the entry content as recognition image information; and Display means for changing each image information of the recognized image information into table display information and screen display information of the recognition display information and displaying the same, and a correspondence relationship in which one screen display information can read out image information related to another screen display information A table information storage means for storing the image information with a table for storing a specific display format readable by the image information in association with the image information for each image information A table memory, when displaying the certain screen display information on the screen of the display means, read out another image information corresponding to the image information from the table information storage means with the image information of the certain screen display information, A display format corresponding to the another image information is output from the display table memory with the read another image information, and the screen display information is displayed in the output display format. A table control unit for giving an instruction to the display unit. 11. The table processing apparatus according to claim 10, further comprising table image storage means for storing said table image information converted by said image input means, wherein said table frame information is stored based on said stored table image information. Extracting means for respectively extracting an item frame and a data frame as table image information corresponding to the data frame, wherein the table image information to be subjected to the character recognition is extracted by the extracted item frame and the extracted data frame And the image information stored in the table information storage means is extracted with the extracted item name, the extracted data frame, the recognition result of the item name, and the entry. A table processing apparatus characterized in that the result is a recognition result of contents. 12. The table processing device according to claim 11, wherein the extracting unit extracts a plurality of ruled lines from the stored table image information, and a plurality of ruled lines surrounded by the extracted ruled lines. Table frame extracting means for extracting an inner frame of the table, item frame extracting means for extracting an item frame of the table information from the extracted inner frame, and an inner table frame adjacent to the extracted item frame. A table processing apparatus comprising: data frame extracting means for extracting a data frame. 13. The table processing device according to claim 12, wherein the item frame extracting means has a length in a direction perpendicular to a direction of a character line of a character described in the table inner frame that is longer than a reference value. A table processing apparatus for extracting a table inner frame adjacent to a long ruled line as the item frame. 14. The table processing apparatus according to claim 13, wherein a length of a table inner frame adjacent to the ruled line in a direction parallel to a direction of the character line is within a range of a reference value. A table processing apparatus, wherein an inner frame is extracted as the item frame. 15. The table processing apparatus according to claim 13, wherein an inner table frame in which an area of the inner table frame adjacent to the ruled line is an area within a range of a reference value is extracted as the item frame. Table processing equipment. 16. The table processing apparatus according to claim 10, wherein the display format is any one of a color display, a mask pattern overlap display, an underline display, a frame surrounding line display, and the like. A table processing device characterized by two displays. 17. The table processing device according to claim 10, wherein, when the certain screen display information is the table display information of the data frame, the another image information is used as the other image information. A table processing apparatus, wherein table image information of an item name belonging to an entry content described in a data frame is used. 18. The table processing device according to claim 10, wherein when the certain screen display information is table display information of the item frame, the another image information is A table processing apparatus, wherein table image information of entry contents belonging to an item name described in an item frame is used. 19. The table processing device according to claim 10, wherein, when the certain screen display information is recognition display information of a recognition result of the entry content, the other image information In the table processing device, wherein the recognition image information of the recognition result of the item name associated with the recognition result of the entry content. 20. The table processing apparatus according to claim 10, wherein, when the certain screen display information is recognition display information of a recognition result of the item name, the different image information Is used as recognition display information of the recognition result of the entry content associated with the recognition result of the item name. 21. The table processing device according to claim 10, wherein the certain screen display information is a table display information of the data frame and a recognition display information of a recognition result of the entry content. Wherein the different image information is recognized image information of a recognition result of an item name associated with the recognition result of the entry. 22. The table processing device according to any one of claims 10 to 16, wherein the certain screen display information is a table display information of the item frame and a recognition display information of a recognition result of the item name. Wherein the another image information is recognized image information of the recognition result of the entry content associated with the recognition result of the item name.