JP4264051B2 - Image processing device - Google Patents

Description

  The present invention relates to an image processing apparatus, and more particularly to accurately separate an image area of an image such as a character in the input image in order to appropriately display the input image, and to display the separated character with high definition. The present invention relates to an image processing apparatus that optimizes image data of a character image.

  An image processing apparatus such as a copying machine or a facsimile machine generates a display image (display image) by performing a process for clearly displaying an input image read from a document, and displays the display image. To do. Various processes are known as the process. For example, when the input image includes characters, a binary image is generated from the input image by slicing each pixel value using a predetermined threshold value and displayed. Alternatively, a pseudo halftone image is generated from the input image by error diffusion / dither processing and displayed. By such processing, the background in the input image is removed, and a clear black character is displayed.

A technique has been proposed in which image area separation processing is performed based on a halftone dot extraction result after reducing the resolution by reducing the input image (see Patent Document 1). Further, a technique for displaying characters with high definition on a color display composed of RGB sub-pixels has been proposed (see Patent Document 2).
JP-A-11-234512 JP 2004-29598 A

  If an image obtained by the various processes described above is viewed as it is on a display device or the like, the following problems may occur. For example, if a binary image obtained by slicing the above-described input image is used, a picture may not be expressed well, and blurring of characters may occur. In addition, when a pseudo halftone image is used, it may be difficult to see due to generation of moire due to display on a display device. Further, when the input image is a color image obtained by reading an old book, the background may be colored because the background is discolored. In addition, when the input image is a color image including characters, the characters are not clear and the images are very difficult to see.

  Further, the various processes described above are processes in an image processing apparatus such as a copying machine or a facsimile machine. That is, the resolution of the input image is high (about 600 dpi) and is fixed. Strictly speaking, it may be a higher resolution, but not a lower resolution. On the other hand, in an image processing apparatus such as a scanner, the resolution takes various values from high resolution to low resolution (about 150 dpi). For this reason, especially in the case of low resolution, the above-described various processes cannot be employed as they are.

  That is, in the various processes described above, image area separation is performed on an image read at high resolution, and appropriate processing is performed on each image area. However, when a large amount of images are captured by a computer, the images are read with low resolution and irreversible compression depending on the data transfer speed from the scanner to the computer and the storage area of the computer. For this reason, in the various processes described above, when there are halftone dots or the like, the image area cannot be accurately separated, and the mosquito noise or block noise of the character part due to lossy compression deteriorates the image quality. In addition, when a document to be read is a document that has passed the years (so-called old books), dust and fading are severe, and the image quality of the input image read by the scanner deteriorates. As a result, when the image is viewed on a display device or the like, it becomes difficult for humans to read. In addition, it is possible to make the background white and the characters black by using a binarized image or the like. However, in this case, since the contrast is too strong, it is difficult to read the image for a long time.

  An object of the present invention is to provide an image processing apparatus that accurately separates an image area of an image such as a character in the input image in order to appropriately display the input image.

  It is another object of the present invention to provide an image processing apparatus that optimizes image data in order to display separated characters with high definition.

An image processing apparatus according to the present invention is an image processing apparatus that performs an image area separation process on an input image, and the resolution of the first reduced image in the input image is determined by the first process based on the characteristics of the image area of a character . A first reduced image creation unit that creates a first reduced image that retains the characteristics of the image area of the character by reducing the input image using a reduction ratio that is smaller than a halftone dot interval ; A reduction rate such that the resolution of the second reduced image is smaller than the halftone dot interval in the input image by a second process different from the first process based on the characteristics of the image area of the pattern. A second reduced image creating unit that creates a second reduced image that retains the characteristics of the image area of the pattern by reducing the input image, and an image area of the character remaining in the first reduced image . Based on the characteristics, the character image area and beyond The determined and the image area, said by determining the image area of the rest and the image area of the picture based on the characteristics of the image area of the second of the picture to remain in the reduced image, a character for the input image, And an image area determination unit that determines which image area is a pattern or background.

According to the image processing device of the present invention , the first reduced image and the second reduced image are created using the reduction ratio described above, and the first reduced image and the image of the pattern that retain the characteristics of the image area of the character. The second reduced image that retains the area characteristics is used to determine whether the image area is a character, a pattern, or a background. Thus, in an image processing apparatus such as a scanner, even when the resolution takes various values from high resolution to low resolution, the image area can be separated with high accuracy even when the resolution is particularly low. Therefore, color burn can be corrected, noise can be removed, smoothed, and color can be restored for each image area that has been accurately determined. As a result, the design and characters can be displayed neatly on the display device. Further, even when the input image is a color image obtained by reading an old book, it is possible to correct the burnt background for easy viewing, display the characters in the color image clearly, and make the image easy to see.

  FIG. 1 is a configuration diagram of an image processing apparatus, and shows an example of the configuration of the image processing apparatus of the present invention. The image processing apparatus 100 stores the input image data (input image) in the input image file 300 and performs predetermined image processing on the input image according to the present invention in order to display the input image neatly on the display device 500. As a result, image data (display image) to be displayed on the display device 500 is generated, stored in the display image file 400, and displayed on the display device 500. The input image is image data read by the image reading apparatus 200 such as the scanner 200 and input to the image processing apparatus 100.

  The image processing apparatus 100 is an image processing apparatus that performs image area separation processing of an input image. In order to perform image processing according to the present invention, an image area processing section 1, a color burn processing section 2, a noise removal section 3, and smoothing are performed. Unit 4, color processing unit 5, and character color processing unit 6. The image area processing unit 1 includes a first reduced image creating unit 11, a second reduced image creating unit 12, and an image area determining unit 13. The color burn processing unit 2 includes a color burn determination unit 21 and a color burn removal unit 22. The color processing unit 5 includes a monochrome / color determination unit 51, a color removal unit 52, and a color restoration unit 53. The character color processing unit 6 includes a brightness optimization unit 61 and a high-definition character processing unit 62.

  The image area processing unit 1 performs an image area separation process on the input image in the image area determination unit 13 for the input image read from the input image file 300. Specifically, the halftone dots of the input image are detected, and based on this detection, the image area of the input image is separated into one of a picture, a character, and a background. Here, the halftone dot includes dither and refers to a color material fixing area on the original during printing (hereinafter the same). For this purpose, the image area processing unit 1 includes a first reduced image creating unit 11 and a second reduced image creating unit 12, and each creates two different reduced images for performing image area separation. The reason for using two different reduced images and the principle of image area determination are as follows.

  Generally, in many originals, a normal pattern is printed with dots called “halftone dots”. The normal picture is, for example, an area (image area) A of “landscape photograph” in FIG. 2 (or an area painted black in FIG. 8). The interval between the halftone dots can be considered to be generally about 100 to 200 lines / inch. Therefore, in the present invention, when creating a reduced image, the resolution of the input image is set so that the resolution of the reduced image is smaller than the interval of halftone dots in the input image (that is, the interval of halftone dots). Change the reduction ratio accordingly. The interval (resolution) sufficiently smaller than the interval between the halftone dots is a resolution at which the halftone dots are crushed, for example, 25 to 75 dpi. In this example, 50 dpi is used (the same applies hereinafter). Thereby, a halftone dot can be detected.

  First, creation of a first reduced image by the first reduced image creating unit 11 will be described. The first reduced image creation unit 11 creates a first reduced image for an input image having a predetermined reduction rate by a first process based on the characteristics of the character image area. The first process is a process for obtaining a reduced image by replacing the target pixel with a pixel having an average value of the pixel values of the target pixel. As described above, the predetermined reduction rate is such a reduction rate that the resolution of the reduced image is smaller than the halftone dot interval in the input image.

  In order to create the first reduced image, the first reduced image creating unit 11 first calculates the average of the values (gradation values) of the respective pixels of the target pixel (processing target pixel) as shown in FIG. Calculate the value. For example, when the resolution of the input image (the image read by the scanner 200) is 600 dpi, as shown in FIG. 3A, in order to set the resolution of the reduced image to 50 dpi as described above, 12 What is necessary is just to obtain | require the average value of the gradation value of a pixel x 12 pixels. Then, the 12 × 12 pixels in the input image are replaced (represented) with the pixels having the average value. Similarly, as shown in FIG. 3B, in the case of an intermediate resolution of 300 dpi, an average value of gradation values of 6 × 6 pixels is obtained. As shown in FIG. 3C, in the case of a low resolution of 200 dpi, an average value of gradation values of 4 × 4 pixels is obtained. As shown in FIG. 3D, in the case of a low resolution of 150 dpi, an average value of gradation values of 3 × 3 pixels is obtained. In the case of other resolutions, an operation close to the resolution described here may be employed. Thereby, a reduced image of 50 dpi is obtained.

  Next, in the obtained reduced image, the difference between the target pixel and each of the surrounding pixels is obtained, the maximum value is selected, and when the selected maximum value (difference) is larger than a predetermined value (threshold), It is determined that the edge component of the character is extracted. The threshold value is set to 20 when the pixel value is expressed by, for example, 256 gradations.

  When a reduced image is created as described above (when the reduced resolution is selected), the halftone dots are crushed but the character edge components remain. Therefore, if the difference from the surrounding pixels is a predetermined value (20 gradations) or more, it can be considered as an edge component of the character. On the other hand, since the pattern originally has few edge components, the edge component disappears when the reduced resolution is selected as described above. The background does not have an edge component. Therefore, by utilizing the difference in presence or absence of the edge component in the reduced image, it is possible to discriminate between the character image area and the other image areas (the image area and the background image area). As a result, as shown in FIG. 2, from an input image in which various image areas are mixed, a halftone dot or image area that strongly includes an edge component (that is, an image area of a character) and another halftone dot or image area ( That is, the image area of the pattern and the image area of the background can be discriminated. Accordingly, the image area of the character using the first reduced image is obtained by obtaining the maximum value in the difference between the pixel of interest and each of the surrounding pixels in the first reduced image, and the maximum value from the predetermined value. If it is larger, it is extracted by determining that the image area of the character has been extracted.

  Next, creation of a second reduced image by the second reduced image creation unit 12 will be described. The second reduced image creating unit 12 creates a second reduced image for the input image having a predetermined reduction rate by a second process different from the first process based on the characteristics of the image area of the pattern. . As described above, the predetermined reduction rate is such a reduction rate that the resolution of the reduced image is smaller than the halftone dot interval in the input image. The second process is a process for obtaining a reduced image by replacing the pixel of interest with a pixel having the minimum pixel value of each pixel of interest calculated by a predetermined process. In the second processing, the minimum value of each pixel value of the target pixel is calculated by different processing according to the resolution of the input image.

  The second reduced image creation unit 12 first creates a minimum value MIN (“black” color) of each pixel value (tone value) of the target pixel (processing target pixel) in order to create a second reduced image. Equivalent to). That is, as a method for creating a reduced image, the image is optimally reduced by taking the minimum value MIN among the gradation values of each pixel of the target pixel when reducing. At this time, the calculation method of the minimum value MIN is changed according to the resolution of the input image.

  In general, when an input image has a high resolution, an example of halftone dots in the image is shown in FIG. High resolution means a resolution at which a halftone dot can be resolved, and is, for example, 600 dpi. In this case, as shown in FIG. 4A, since the halftone dots (in the region A in FIG. 2) can be clearly resolved (read), black components (halftone components) can be extracted. When the input image has a low resolution, an example of halftone dots in the image is shown in FIG. Low resolution refers to a resolution at which halftone dots cannot be resolved, and is, for example, 150 dpi. In this case, as shown in FIG. 4B, since the halftone dots (in the area A in FIG. 2) cannot be resolved (cannot be read), variations occur in the extraction of the halftone line segments.

  From the above, when the input image has a high resolution, the minimum value MIN may be taken from the target pixel group as shown in FIG. That is, for example, a pixel having the smallest value is selected from 4 × 4 = 16 target pixels (values) d11 to d44, the value is set to the minimum value MIN, and the 16 target pixels d11 to d11 are selected by the pixel s11. Replace d44. That is, the image is reduced by representing the pixels of interest d11 to d44 with the pixel s11. Thereby, a high resolution halftone dot component (black component) is extracted.

  On the other hand, when the input image has a low resolution, as shown in FIG. 5B, for example, a 2 × 2 region is obtained by performing a predetermined calculation on 4 × 4 = 16 target pixel groups. Extract black component. That is, by using the four target pixels d11, d12, d21, and d22 among the 16 pixels, the temporary black component t11 = 255 − ((255−d11) + (255) by the calculation shown in FIG. -D12) + (255-d21) + (255-d22)). Similarly, other black components t12, t21, and t22 are calculated. The pixel with the smallest value is selected from these four extracted temporary black components, the value is set to the minimum value MIN, and the 16 pixels of interest d11 to d44 are replaced with the pixel s11 to reduce the image. To do. Thus, a low resolution halftone dot component (black component) is extracted while suppressing variation due to halftone dots.

  For example, when the resolution of the input image is 600 dpi, as shown in FIG. 6A, in order to set the resolution of the reduced image to 50 dpi, the gradation value is selected from 12 × 12 pixels. The smallest value, that is, the minimum value MIN is obtained by the method shown in FIG. The minimum value MIN is, for example, about 20 to 30 with 256 gradations. Then, the 12 × 12 pixels in the input image are replaced with the pixel having the minimum value MIN. Similarly, as shown in FIG. 6B, in the case of an intermediate resolution of 300 dpi, the minimum value MIN is obtained from 6 × 6 pixels by the method shown in FIG. In the case of the low resolution of 200 dpi shown in FIG. 6C and the low resolution of 150 dpi shown in FIG. 6D, the minimum value is selected from 4 × 4 pixels and 3 × 3 pixels, respectively. The MIN is obtained by the method shown in FIG. In the case of other resolutions, an operation close to the resolution described here may be employed. Thereby, a reduced image of 50 dpi is obtained.

  Next, a histogram is created for the obtained reduced image as shown in FIG. Using this histogram, a threshold value T based on variance or the like is determined, and the reduced image is binarized using this threshold value T. The threshold T is calculated by, for example, the calculation shown in FIG. As a result, a black area having a certain area is generated in the input image with respect to the halftone dot or the image area of the pattern.

  In this case, the halftone dots and characters are further distinguished as follows. For example, the size of a 12-point character is about 4 × 4 mm, and on the other hand, a gap is generated in the character and black and white components are formed. Considering these, when the proportion of the black component in a region of 10 × 10 dots (5 mm × 5 mm) is high, it can be determined that the region is not a character but a dot or a picture. .

  In general, the black component of the pattern is larger than the black component of the character, and the black component of the character disappears when the reduced resolution is selected as described above. On the other hand, for the picture, the black component remains because the black component is acquired from the adjacent pixel even at low resolution. The background does not have an edge component. Accordingly, by utilizing the difference in the ratio of the black component (black pixel) in the reduced image, the image area of the picture having a large amount of black component and the other image area (that is, the character image area and the background image area). ). Therefore, if there are, for example, 98 or more black components (black pixels) in 10 × 10 = 100 pixels in the reduced image, it is determined that the black component of the pattern has been extracted. As a result, as shown in FIG. 8, from an input image in which various image areas are mixed, an image area in which many black components exist (that is, an image area of a pattern) and an image area other than this (that is, an image of a character). Area and background image area). Therefore, the image area of the pattern using the second reduced image is binarized from the second reduced image by a value determined from the histogram created based on the second reduced image, and the black pixel in the binarized image is obtained. When the ratio is larger than a predetermined value, it is extracted by determining that the image area of the pattern is extracted.

  As described above, by comparing the two reduced images shown in FIGS. 8 and 2, the image area of the pattern, characters, and background can be determined. Therefore, the image area determination unit 13 obtains a reduced image with a reduction ratio corresponding to the resolution of the input image, extracts the image area of the character using the first reduced image, and uses the second reduced image to design the picture. The image area is extracted, and the background image area is extracted using the first reduced image and the second reduced image, thereby determining whether the input image is a character, a picture, or the background image area. To do. In addition, by creating two reduced images, even when the input image has a low resolution, it is possible to perform high-accuracy image area separation, and to improve accuracy thereafter. Furthermore, by extracting the image area of the character in the pattern and performing the optimum processing according to the image area, it is possible to make the character of the image easy to read. As a method for obtaining a reduced image, for example, a well-known reduction method such as bilinear / bicubic may be used.

  The color burn determination unit 21 determines the presence or absence of color burn. That is, the image area processing unit 1 determines whether or not the area determined to be the background image area is a color burn image. In general, as shown in FIG. 9A, a color-burned document has a feature that the red component R 赤 of the background is strong and the blue component B￣ is weak. FIG. 9A is a histogram of the background image area. Further, if the tendency of R￣ <G￣ <B￣ is strong, the tendency of uneven color is strong. Therefore, in the image area determined as the background by the above-described image area determination, pixels satisfying R> (G + 10)> (B + 20) at positions corresponding to each other are compared with the total number of pixels in the background image area. If the ratio is high, it is determined that the document is a color burned document. For example, R is the red component value (gradation value) of the pixel, and R￣ is the average value (gradation value) of the red component of each pixel in the image area (background image area). Yes (the same is true for others). Furthermore, the color fading value K used in the color restoration unit 53 described later is obtained by K = R−B.

  For example, the color burn removing unit 22 removes color burn from an area that is a background image area (in the input image) that is determined to be a color burn image by the color burn determining unit 21. When it is determined that the document is a color-burned original, the color-burn removal unit 22 matches the RGB values in each pixel as shown in FIG. 9B (makes the RGB histograms overlap). That is, the color burn removal unit 22 creates a histogram for each RGB component as shown in FIG. 9A for the pixels in the determined background image area. The value calculated from the average value or the variance value of this histogram is white (255 in pixel gradation values (in 256 gradations)). That is, G and B tone curves are changed with R as a reference. Thereby, as shown in FIG. 9B, the color burn is removed.

  Therefore, according to the present invention, it is determined whether or not the image area of the background that has been accurately discriminated is faded (color fading). If the image area is burnt, unnecessary colors can be removed from the background. Thus, for example, when the input document is an old book, the background image is removed by determining the color burn specific to the old book by paying attention to the separated background image area, and other parts as described later. The contrast of the image can be increased to make the image easier to see.

  The noise removing unit 3 removes noise from the input image. In general, in a color-burned document, deterioration of an image obtained from an input device is observed due to deterioration due to the passage of time, adhesion of dust, or the like. Therefore, it is necessary to remove dust. As shown in FIG. 10, the noise removing unit 3 further binarizes the reduced image created by the image area processing unit 1. At this time, when the reduced image is binarized, the dust read as shown on the left side of FIG. 10 normally does not have a continuous area around it, and thus becomes an isolated point in the image. Therefore, a large noise component is removed by removing this isolated point. The dust that can be removed by the noise removing unit 3 is one dot at a resolution of 50 dpi, that is, dust that is smaller than about 0.5 mm. As described above, since the size of the 12-point character is about 4 mm, the character is not erased.

  The smoothing unit 4 smoothes the image (input image) for each image area. That is, smoothing is performed by different filters for each of the regions determined to be the character, pattern, and background image regions in the input image. For this purpose, the smoothing unit 4 is prepared for each image area, for example. That is, an optimum filter for smoothing is prepared for each image area separated as described above (an optimum smoothing unit 4 is prepared). Thus, the optimum noise component removal processing is performed for each image area.

  In general, in the background image area, there may be block noise due to lossy compression, small dust / scratches, and the like. In the image area of the pattern, there may be variations in moire and resolution depending on the halftone dots and the characteristics of the input device. In addition, since the edge component is strong in the character image area, mosquito noise due to irreversible compression may occur.

  In the background image area, basically, as shown in FIG. 11A, the smoothing filter is weakened to remove the noise component. That is, a filter with almost no blur is used. On the other hand, looking at block noise in the background image area, noise is generated at the boundary of an 8 × 8 area (a well-known JPEG DCT block) due to lossy compression. Therefore, as shown in FIG. 11B, the smoothing filter is strengthened to remove the noise component at the boundary of the 8 × 8 region. In other words, a slightly strongly blurred filter is used.

  In the image area of the pattern, in order to remove halftone moiré and resolution variations, the smoothing filter is strengthened to remove noise components (halftone dots) as shown in FIG. 11C. . That is, a strongly blurred filter is used.

  In the character image area, mosquito noise due to lossy compression is larger as the edge component is stronger in the 8 × 8 area as shown in FIG. Therefore, the difference between the target pixel and the surrounding pixels is taken, and smoothing is performed with the pixel when the difference is smaller than a certain threshold value. Specifically, as shown in FIG. 12B, a noise component is removed using a special filter. That is, the threshold value is increased as the edge component increases in the 8 × 8 region. As a result, the character component is smoothed by the character component, and the other portions are the other portions, and the noise components can be removed by performing smoothing, respectively.

  Here, depending on variations in the strength of the smoothing filter for each image area, an edge component (step of gradation value) may be formed at the boundary of the image area. Therefore, as shown in FIG. 12C, a smoothing filter intermediate between the image area and the image area is used at the boundary of the image area. The filter shown in FIG. 12C is an intermediate smoothing filter used at the boundary between, for example, a pattern image area and a background image area. This may prevent the edge component from being noticeable.

  As described above, the noise component can be removed from the input image, and the smoothing process according to the image area can be performed. In particular, the smoothing filter is changed for each image area accurately determined as described above, thereby smoothing the image, background noise, halftone dots in the picture area, mosquito noise due to lossy compression of the character area, etc. In addition, the data obtained by reading dust / dust on the original can be smoothed to make the image clean. In addition, noise peculiar to the image reading apparatus 200, block noise / mosquito noise peculiar to lossy compression, dust peculiar to old books, and the like can be deleted.

  The monochrome / color determination unit 51 determines whether the input image is a monochrome image or a color image. In general, some originals are not always black characters on a white background. That is, since the background color of the document may be a color, it is necessary to make a determination. Therefore, the monochrome / color determination unit 51 extracts a color component for an image area other than the background image area. In this example, for each pixel, the difference between the maximum value Max and the minimum value Min in RGB is extracted as a color component. At this time, instead of the RGB space, the color space may be converted into HSV saturation or Lab ab value, and the value may be used as a color component. The monochrome / color determination unit 51 determines that the image is a color image when the value of the obtained color component is larger than a predetermined value.

  For example, when it is determined that the input image is a monochrome image, the color removal unit 52 removes a color (color component) from the input image. That is, when it is determined that the input document is a monochrome image, the color removal unit 52 removes the color component because it is useless. As a removal method, the value of the pixel may be converted from RGB to gray (gray) by obtaining Y as a value Y by Y = 0.30R + 0.59G + 0.11B. At this time, it may be converted into HSV or Lab known in the color space, and after removing the color component, it may be converted back to RGB and grayed.

  For example, when it is determined that the input image is a color image, the color restoration unit 53 restores the color (color component) of the input image. That is, when it is determined that the input document is a color image, the color restoration unit 53 performs a process of restoring the color. Generally, it is considered that the color burned manuscript is also faded. Therefore, the saturation is improved in accordance with the degree of fading calculated by the color burn determination unit 21 (color fading value K, see FIG. 9A). Specifically, the RGB component is converted from HSV to HSV, and the S component of the HSV is increased with respect to K as the faded value K increases. The color is restored by returning the HSV calculated in this way to RGB.

  As described above, when it is determined that the image is a color image, color restoration is performed (the color of the faded document is restored), and when it is determined that the image is monochrome, the color component is removed to achieve the optimum contrast depending on the brightness. Can do. Thereby, for example, an input image obtained by reading an old book or the like can be displayed neatly on the display device 500 and can be easily viewed.

  The brightness optimization unit 61 optimizes the brightness of the input image. That is, processing is performed to make the characters black and the background white. In general, when a character or the like is viewed on the display device 500, the higher the contrast, the clearer it looks. Therefore, the white level and the black level are determined so that the characters are black and the background is white. For example, the brightness optimization unit 61 creates a histogram for an image from which color has been removed, and obtains a predetermined value (slice) based on the variance value of the histogram. Further, the maximum value Max and the minimum value Min of the histogram are obtained. An intermediate point (gradation) between these slices and the minimum value Min is black, and an intermediate point between the slices and the maximum value Max is white. Therefore, the background can be white and the characters can be black. As a result, it is possible to reduce blurring and jaggy of the character image area, and the image area of the pattern is easy to see on the display device 500 or the like because it has gradation.

  The high-definition character processing unit 62 performs high-definition processing on at least a character image area for an input image including characters. That is, the high-definition character processing unit 62 further performs high-definition processing of characters included in the input image so that the display device 500 can easily see the character. That is, as shown in FIG. 13, the high-definition character processing unit 62 assumes that RGB of each pixel is continuous, and performs high-definition character processing as will be described later with reference to FIG. That is, for at least an area determined to be a character image area, after converting characters included in the input image into high-definition characters of a monochrome image, the high-definition characters of the monochrome image are converted into high-definition characters of a color image. As a result, the character is thinned and the periphery of the character is slightly colored. Thereby, an LCD (liquid crystal display device) or the like has an effect of lighting a sub-pixel, and jaggy reduction is achieved. In addition, the color of the character is slightly contrasted so that the contrast of the character is clear.

  Furthermore, a white character (pixel value (tone value) is 255) and a black character (pixel value is 0) has a high contrast (difference between the white background and the black character) and is easy to see in a short time. However, when looking at the display device 500 for a long time, the contrast becomes too high, and the eyes become tired. Therefore, the pixel value is slightly added to the black character to reduce the contrast and make it easy to read. That is, the contrast of the image is reduced by setting the gradation value of the black character pixel to a value larger than the original value. Specifically, the high-definition character processing unit 62 sets the gradation value of the pixel of the black character to a value larger than 0, for example, 20 (256 gradations). At this time, RGB may be converted into Lab, and the color difference ΔE (in this case, brightness) may be set to a gradation value of 80 (256 gradations), for example, and returned from Lab to RGB.

  Thereafter, the high-definition character processing unit 62 stores the image obtained by the above processing in the display image file 400 as an output image. The output image is displayed on the display device 500 by the image processing apparatus 100. At this time, according to the present invention, a clearer and easier-to-read image is displayed on the display device 500 than when the document or the input image is directly displayed.

  As described above, jaggies can be reduced and contrast can be improved by high-definition processing of characters. Further, in order to make the image easy to read on the display device 500 or the like, it is possible to display an easy-to-read image even for a long time by performing optimization based on brightness and filtering processing for characters. Further, the high-definition character processing unit 62 can display an easy-to-read image with high contrast and without eye fatigue on the display device 500 by using emphasis and blurring together.

  As described above, according to the present invention, high-precision image area separation is possible even at low resolution, filter processing corresponding to each separated image area is performed, and the display device 500 and the like are easy to see. It is possible to make the image easier to read by performing the processing for the purpose.

  FIG. 14 is an image processing flow and shows an example of image processing in the image processing apparatus 100 of the present invention. The image area processing unit 1 performs image area determination processing (step S11). The image area determination process will be described later with reference to FIGS.

  Thereafter, the color burn processing unit 2 performs color burn determination processing by the color burn determination unit 21 (step S12), and checks the presence or absence of color burn based on the determination result (step S13). When there is color burn, the color burn removing unit 22 performs color burn removal processing (step S14), and when there is no color burn, step S14 is omitted. Thereafter, the noise removing unit 3 performs a noise removing process (step S15), and the smoothing unit 4 performs a smoothing process optimum for the image area for each image area determined in step S11 (step S16). Note that steps S15 and S16 may be executed in parallel. Thereafter, the color processing unit 5 performs monochrome / color determination processing for the image by the monochrome / color determination unit 51 (step S17), and checks whether the image is a monochrome image or a color image based on the determination result (step S18). ). When the image is a monochrome image, the color removal unit 52 performs a color removal process (step S19), and when the image is a color image, the color restoration unit 53 performs a color restoration process (step S110).

  Thereafter, the character color processing unit 6 performs lightness optimization processing by the lightness optimization unit 61 (step S111), and performs high-definition character processing by the high-definition character processing unit 62 (step S112). High-definition character processing will be described later with reference to FIGS.

  FIG. 15 is a reduced image creation process flow, and shows an example of a reduced image creation process in the image area determination process in the image processing apparatus 100 of the present invention. This process is executed by the first reduced image creating unit 11 in step S11 of FIG.

  The first reduced image creating unit 11 creates the first reduced image as described above (step S21), takes out one target pixel from the first reduced image, and extracts the target pixel and a plurality of surrounding pixels. The difference between each pixel (a plurality of surrounding pixels) is obtained, the maximum difference among them is extracted (step S22), and whether the maximum difference is a value equal to or higher than a predetermined level (for example, gradation difference 20). (Step S23). When the maximum difference is equal to or greater than a predetermined level, the first reduced image creation unit 11 determines that the target pixel is an edge (edge region) (step S24). This means that the image area of the character is determined as described above. If the maximum difference is not greater than or equal to the predetermined level, it is determined that the pixel of interest is another region (step S25). As described above, this means that an image area other than characters (image and background image areas) is determined. Thereafter, the first reduced image creating unit 11 checks whether or not the processing for all the pixels has been completed (step S26), and if not completed, repeats step S22 and subsequent steps. It is determined that the pixel is also an edge (step S27). Thereby, the result on the right side of FIG. 2 is obtained.

  FIG. 16 is a reduced image creation process flow, and shows an example of a reduced image creation process in the image area determination process in the image processing apparatus 100 of the present invention. This process is executed by the second reduced image creating unit 12 in step S11 of FIG. 14 in parallel with or after the process of FIG.

  The second reduced image creating unit 12 creates the second reduced image as described above (step S31), creates a histogram for the second reduced image (pixel value thereof) (step S32), and Based on the histogram, the second reduced image is binarized to obtain a binary image (step S33). Thereafter, the second reduced image creating unit 12 counts the number of black pixels in the binary image (step S34), and black pixels of a predetermined level (number) or more are detected (in 100 pixels). It is checked whether 98 or more have been detected (step S35). Usually, the image area of the pattern is black. The second reduced image creating unit 12 determines that the pixel is a halftone pixel (picture) when a black pixel of a predetermined level (number) or more is detected (step S36). This means that the image area of the pattern is determined. If a black pixel of a predetermined level or higher is not detected, it is determined that it is another pixel (step S37). This means that an image area (character and background image areas) other than the pattern is determined. Thereafter, the second reduced image creating unit 12 checks whether or not the processing for all the pixels has been completed (step S38), and if not completed, repeats step S34 and subsequent steps, and if completed, ends the processing. To do. Thereby, the result shown on the right side of FIG. 8 is obtained.

  FIG. 17 is an image area determination process flow, and shows an example of an image area determination process in the image processing apparatus 100 of the present invention. This processing is executed by the image area determination unit 13 after the processing in FIGS. 15 and 16 in step S11 in FIG.

  The image area determination unit 13 checks whether or not the result of the process using the second reduced image (the process in FIG. 16) indicates that the pixel area is a picture area (step S41), and the process result Indicates that it is a picture area, it is determined that the area is a picture area (image area, hereinafter the same) (step S42). When the processing result does not indicate that the image area is a picture area, the image area determination unit 13 indicates that the result of the process using the first reduced image (the process in FIG. 15) is that the pixel area is a character area. Whether or not it is indicated is checked (step S43), and if the processing result indicates that it is a character area, it is determined that the area is a character area (step S44). If the processing result does not indicate that the area is a character area, the image area determination unit 13 determines that the area is a background area (step S45). Thereafter, the image area determination unit 13 checks whether or not the processing for all the pixels has been completed (step S46). If the processing has not been completed, step S41 and subsequent steps are repeated. If the processing has been completed, the processing is terminated.

  FIG. 18 is a high-definition character processing flow, and shows an example of high-definition character processing in the image processing apparatus 100 of the present invention. This process is executed by the high-definition character processing unit 62 in step S112 of FIG.

  The high-definition character processing unit 62 performs a process of converting a character included in the input image into a high-definition character of a monochrome image (step S51), and checks whether or not the process has been completed for all pixels (step S52). If not completed, step S51 and subsequent steps are repeated. When the processing is completed, the high-definition character processing unit 62 thereafter performs a process of converting the high-definition characters of the monochrome image into the high-definition characters of the color image, and thereby, around the black character based on the surrounding color A fixed color is assigned (steps S53 to S55), and it is checked whether or not the processing has been completed for all the pixels (step S56). If not, the processing from step S53 is repeated. To do. The high-definition character processing of the color image includes processing for the R component in the color determined based on surrounding colors (step S53), processing for the G component (step S54), and processing for the B component (step S55). It is executed in order.

  Note that in steps S52 and S56, the pixels (characters) to be processed are at least all the pixels in the range defined as the character image area by the image area determination described above. Therefore, the processing of FIG. 18 may be performed for all image areas of the input image.

  FIG. 19 is a high-definition character processing flow, and shows an example of processing for the R component in the high-definition character processing in the image processing apparatus 100 of the present invention. This process is executed by the high-definition character processing unit 62 in step S53 of FIG.

  The high-definition character processing unit 62 obtains the parameter Alfa (step S61), obtains the parameter Beta based on the parameter Alfa (step S62), and obtains the first intermediate value Tmp1 using these parameters (step S63). The first intermediate value Tmp1 is a weight calculated based on the pixel on the left side (or the right side) of the pixel at the edge of the character. In addition, the value of each pixel shall be represented by 256 gradations of 0-255. Similarly, the high-definition character processing unit 62 obtains the parameter Alfa (step S64), obtains the parameter Beta (step S65), and obtains the second intermediate value Tmp2 (step S66). The second intermediate value Tmp2 is a weight calculated based on the pixel on the right side (or the left side) of the pixel at the edge of the character. Thereafter, the high-definition character processing unit 62 calculates the R component value r [i] by averaging the first intermediate value Tmp1 and the second intermediate value Tmp2 (step S67).

  20 and 21 are high-definition character processing flows, and show an example of processing for the G component and B component in the high-definition character processing in the image processing apparatus 100 of the present invention. These processes are respectively executed by the high-definition character processing unit 62 in steps S54 and S55 in FIG. 18 and are both similar to the processes in FIG.

  As described above, according to the present invention, even when the resolution of the input image takes various values from high resolution to low resolution, the image area can be separated with high accuracy in the image processing apparatus. Various processes such as correction of color burn can be performed for each image area. As a result, the design and characters can be displayed neatly on the display device. Further, even when the input image is a color image obtained by reading an old book, it is possible to correct the burnt background for easy viewing, display the characters in the color image clearly, and make the image easy to see.

  According to the present invention, in the image processing apparatus, it is possible to compensate for the deterioration of the image quality of the character image area due to various reasons, and it is possible to display the characters neatly on the display device. Can be displayed neatly on the display device.

It is an image processing apparatus block diagram. It is image area determination processing explanatory drawing. It is image area determination processing explanatory drawing. It is image area determination processing explanatory drawing. It is image area determination processing explanatory drawing. It is image area determination processing explanatory drawing. It is image area determination processing explanatory drawing. It is image area determination processing explanatory drawing. It is color processing explanatory drawing. It is noise removal explanatory drawing. It is smoothing explanatory drawing. It is smoothing explanatory drawing. It is high definition explanatory drawing. It is an image processing flow. It is a reduced image creation processing flow. It is a reduced image creation processing flow. It is an image area determination processing flow. This is a high-definition character processing flow. This is a high-definition character processing flow. This is a high-definition character processing flow. This is a high-definition character processing flow.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Image area process part 2 Color burn process part 3 Noise removal part 4 Smoothing part 5 Color process part 6 Character color process part 11 1st reduction image creation part 12 2nd reduction image creation part 13 Image area determination part 21 Color burn determination part 22 Color burn removal unit 51 Monochrome / color determination unit 52 Color removal unit 53 Color restoration unit 61 Brightness optimization unit 62 High-definition character processing unit

Claims (12)

An image processing apparatus that performs image area separation processing of an input image,
The input image is reduced using a reduction ratio such that the resolution of the first reduced image is smaller than the halftone dot interval in the input image by the first processing based on the characteristics of the character image area. A first reduced image creating unit for creating a first reduced image that retains the characteristics of the image area of the character ;
A reduction ratio is used such that the resolution of the second reduced image is smaller than the halftone dot interval in the input image by a second process different from the first process based on the characteristics of the image area of the pattern. A second reduced image creating unit that creates a second reduced image that retains the characteristics of the image area of the pattern by reducing the input image ,
Based on the characteristics of the image area of the character remaining in the first reduced image , the image area of the character and the other image area are distinguished, and the image area of the pattern remaining in the second reduced image is characterized. An image area determination unit that determines whether the input image is an image area of a character, a pattern, or a background by determining an image area of the pattern and an image area other than the image area based on the image area; An image processing apparatus. The first process is a process of obtaining a reduced image by replacing the target pixel with a pixel having an average value of the pixel values of the target pixel.
The second process is a process of obtaining a reduced image by replacing a target pixel with a pixel having a minimum value of each pixel value of the target pixel calculated by a predetermined process. The image processing apparatus described. The image processing apparatus according to claim 2, wherein, in the second process, a minimum value of each pixel value of the target pixel is calculated by a process different according to the resolution of the input image. In the image area of the character using the first reduced image, the maximum value in the difference between the pixel of interest and each of the surrounding pixels is obtained in the first reduced image, and the maximum value is greater than a predetermined value. The image processing apparatus according to claim 1, wherein the image processing device is extracted by determining that the image area of the character has been extracted. The image area of the pattern using the second reduced image is binarized from the second reduced image by a value determined from a histogram created based on the second reduced image, and black pixels in the binarized image are obtained. The image processing apparatus according to claim 1, wherein the image processing device is extracted when it is determined that the image area of the pattern is extracted when the ratio is larger than a predetermined value. The image processing apparatus further includes:
For the area determined to be the background image area, in the background image area, the red component is larger than the sum of the green component and the first predetermined value, and the green component and the first predetermined value are Whether the image is a color burn image by determining that it is a color burn image when the ratio of the number of pixels greater than the sum of the blue component and the second predetermined value is higher than the total number of pixels in the image area A color burn determination unit for determining whether or not,
The image processing apparatus according to claim 1, further comprising: a color burn removing unit that removes color burn from an area that is a background image area determined to be a color burn image. The image processing apparatus further includes:
2. The image according to claim 1, further comprising: a plurality of smoothing units that perform smoothing using different filters for each of the regions determined to be image regions of characters, designs, and backgrounds in the input image. Processing equipment. The image processing apparatus further includes:
A monochrome / color determination unit for determining whether the input image is a monochrome image or a color image;
When it is determined that the input image is a monochrome image, a color removal unit that removes a color component from the input image;
When it is determined that the input image is a color image, for each pixel of the color image, the saturation of the pixel is determined according to a color fading value that is a value obtained by subtracting a blue component value from a red component value. The image processing apparatus according to claim 1 , further comprising a color restoration unit that restores a color component of the input image by improving the color component. The image processing apparatus further includes:
For at least the area determined to be the character image area, after converting characters included in the input image into high-definition characters of a monochrome image, the high-definition characters of the monochrome image are converted into high-definition characters of a color image. The image processing apparatus according to claim 1, further comprising a high-definition character processing unit. The image processing apparatus further includes:
For the area determined to be the background image area, a color burn determination unit that determines whether the image is a color burn image;
A color burn removal unit that removes color burn from an area that is a background image area determined to be a color burn image;
A plurality of smoothing units that perform smoothing by different filters for each of the regions determined to be image regions of characters, designs, and backgrounds in the input image;
A monochrome / color determination unit for determining whether the input image is a monochrome image or a color image;
When it is determined that the input image is a monochrome image, a color removal unit that removes a color component from the input image;
When it is determined that the input image is a color image, a color restoration unit that restores a color component of the input image;
For at least the area determined to be the character image area, after converting characters included in the input image into high-definition characters of a monochrome image, the high-definition characters of the monochrome image are converted into high-definition characters of a color image. The image processing apparatus according to claim 1, further comprising a high-definition character processing unit. The image processing apparatus further includes:
The image processing apparatus according to claim 10 , further comprising a brightness optimization unit configured to make the characters black and the background white. The image processing apparatus according to claim 10, wherein the high-definition character processing unit reduces the contrast of the image by setting a gradation value of a pixel of a black character to a value larger than an original value.

Images