JP2001136532A - Encoding device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To generate bit stream of only arbitrary object from among existent bit streams, without deteriorating the image. SOLUTION: An object extraction part extracts a specified object from an image obtained by decoding an existent bit stream by a decoding part. An object position information part generates position information on the extracted object and outputs it to a bit stream editing part. The bit stream editing part generates bit streams in this specified object units by using the existent bit stream as much as possible but generates the bit stream in the mentioned object units, by using a bit stream which is decoded once and then re-encoded, only when information on macroblocks other than the object needs to be referred to. Since almost all bit streams in the object units use the existent bit stream which is not re-encoded, deterioration of image can be eliminated.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an encoding apparatus that creates a shape-encoded bit stream of only a specific object specified separately from an existing encoded bit stream.

[0002]

2. Description of the Related Art Conventionally, when an existing bit stream represented by, for example, MPEG (High Efficiency Coding) is decoded, one object is extracted from the decoded bit stream, and only a macro block constituting the portion is extracted. Identifying and creating a bit stream composed of only these macro blocks is performed.

FIG. 5 is a block diagram showing a method of creating a bit stream of only a specific object from a conventional coded bit stream 100. After the existing bit stream 100 is decoded by the decoding unit 11, an object specified separately is extracted by the object extraction unit 12, and image data 200 of the object is generated.

[0004] The image data 200 of the object is encoded by the object encoding unit 13 to generate a bit stream 300 whose shape is encoded for each object.

In such a conventional method, in order to create a stream of only a specific object from a bit stream that has already been created, it is necessary to decode once and then perform encoding for generating an object again. I had to.

[0006]

In the above-described conventional method of creating a shape-encoded bit stream of only a specific object from an existing encoded bit stream, an object is extracted from the existing bit stream. When performing encoding by decoding, an object must be specified by decoding, and the object must be re-encoded. Therefore, there is a problem that the image is deteriorated by the re-encoding.

SUMMARY OF THE INVENTION The present invention has been made to solve the conventional problems as described above, and has as its object to provide an image processing apparatus which does not degrade an image and designates an arbitrary specified bit stream from an existing coded bit stream. An object of the present invention is to provide an encoding device capable of creating an object shape encoded bit stream.

[0008]

In order to achieve the above object, a feature of the present invention is that a bit stream comprising only an arbitrarily designated object from among existing coded bit streams. Means for extracting blocks forming the object in the existing coded bit stream, and forming a shape-encoded bit stream using only the extracted blocks. Is provided.

According to the first aspect of the present invention, when the blocks constituting the object are generated with reference to only the image in the object, the shape encoding is performed only by the existing encoded bit stream. Because
Good image quality can be obtained.

A second aspect of the present invention is the encoding apparatus according to the first aspect, wherein the position information of the object indicates that a macroblock constituting the object refers to an image outside the object. ,
The object of the present invention is to construct a shape-encoded bit stream using blocks obtained by re-encoding corresponding image blocks.

According to the second aspect of the present invention, if a block constituting the object is generated by referring to an image outside the object, only the block is re-created. Since it is rare to shape-encode using the one obtained by encoding and use the block obtained by re-encoding, there is almost no image deterioration.

[0013] A feature of the invention according to claim 3 is that decoding means for decoding an existing encoded bit stream to obtain image data, and an outline of an object separately designated from the image data obtained by the decoding means. Object position information generating means for generating a shape signal in units of macroblocks corresponding to the object from the outline information of the object obtained by the object extracting means, and the object position information generating means. When it is determined based on the obtained shape signal that the macroblocks constituting the object are generated with reference to only the image in the object, the object is constructed using the existing bit stream. Macro block is a picture outside the object And a bit stream editing unit that creates a shape-encoded bit stream using a bit stream obtained by re-encoding the image data obtained by the decoding unit when it is determined that the bit stream is generated by referring to Is to do.

According to the third aspect of the present invention, when the blocks constituting the object are generated with reference to only the image in the object, the shape encoding is performed only with the existing encoded bit stream. If any of the blocks constituting the object is generated with reference to an image outside the object, only the corresponding block is subjected to shape coding using a block obtained by re-coding the block of the corresponding image. Moreover, since it is rare to use a block obtained by re-encoding, it can be said that most of the objects are shape-encoded only with an existing encoded bit stream, and image quality does not decrease, Good quality images can be obtained.

[0014]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing an embodiment of the encoding device of the present invention. The encoding device is
For example, a decoding unit 1 for decoding an existing bit stream 100 represented by MPEG, from among image data,
For example, an object extracting unit 2 for extracting an outline of an object specified by a user, an object for generating position information (shape signal 60 for each macroblock) from the outline information of the extracted object, and outputting the position information to a bit stream editing unit 4 It comprises a position information section 3 and a bit stream editing section 4 for editing an existing bit stream 100 and a decoded bit stream to create a bit stream (shape coded bit stream) 300 in object units.

FIG. 2 is a block diagram showing a detailed configuration example of the bit stream editing unit 4 shown in FIG. The bit stream editing unit 4 reconstructs a motion vector to obtain a true vector, a target macroblock information extracting unit 41,
The vector range determination unit 42 that determines whether the target macroblock refers to an image in a macroblock that constitutes the object, without omission (no skip)
A target block information reconstructing unit 43 for reconstructing a complete macroblock, selecting one of macroblock information from the target block information reconstructing unit 43 and macroblock information from the shape encoding unit 45 It has a switch unit 44 for outputting, a shape encoding unit 45 for re-encoding the decoded image, and a shape-encoded bit stream creating unit 46 for creating a bit stream for each specified object.

Next, the operation of this embodiment will be described. Existing bitstreams 100 typically have a so-called M
It is encoded by PEG1 or MPEG2. On the other hand,
There is image transmission of an arbitrary shape different from conventional rectangular image transmission suitable for low-rate transmission. There is no problem if these arbitrary-shaped images are originally encoded as arbitrary-shaped images.However, in order to create a required arbitrary-shaped image from an existing rectangular image, conventionally, re-encoding is performed. Was solved.

Extracting only a part of the objects from these rectangular images into an arbitrary-shaped image is performed by storing a bit stream to be an arbitrary-shaped image in a database and editing an image combining these. Is very important for the processing system.

Therefore, the existing bit stream 100 is decoded by the decoding unit 1 (see FIG. 1), and the image data 80 is decoded.
Then, the object is specified. The user specifies the object while viewing the decoded image, for example. For the designated object, the transect of the object is extracted by the object extracting unit 2, and the outline information is input to the object position information unit 3.

The object position information section 3 generates a shape signal 60 for each macroblock from the contour information, and outputs it to the bit stream editing section 4. On the other hand, both the existing bitstream 100 and the image data 80 decoded by the decoding unit 1 are sent to the bitstream editing unit 4.

The bit stream editing unit 4 specifies a target macroblock and a block included in the macroblock based on the shape signal 60 from the object position information unit 3, and inputs the input shape signal 60 to the macroblock. The luminance signal and the color difference signal are output. It should be noted that the transmission signal used in the actual display can be held until the display of each object is determined, so that it is not considered here.

Here, the existing bit stream 100 and the decoded image data 8 are stored in the bit stream editing unit 4.
The reason why 0 is sent and edited together will be described below.

In shape coding, an image block includes, for example, as shown in FIG. 3, an entire macroblock (square frame) included in an object (hatched portion) or a boundary between an object and a portion other than the object. Macroblocks are classified into three types: macroblocks, and macroblocks that do not contain objects (here as backgrounds).

In this case, the transform blocks required for the shape encoding are blocks in which the entire block is included in the object and blocks including the boundaries of the object. These blocks correspond to so-called MPEG1 or MPEG2 macroblocks. Therefore, these macroblocks have already been encoded.

However, in shape-encoded coding from the beginning, the reference image of a block referring to another frame always exists in a set of transform blocks necessary for coding this object, or It can be reconstructed using the data in this object.

However, a block coded by MPEG1 or MPEG2 uses a reference image in a range defined by an object of a previous frame (or a subsequent frame) as shown in FIG. In rare cases, the macroblock of the encoding target frame as shown in (B) is not always reconstructed. This is a one-way prediction,
The same is true for bidirectional prediction.

When such a rare case occurs, since there is no reference image, using the macro block as it is causes a problem. If the predicted condition is the same as that of the macroblock in the previous frame, coding may not be performed. In this case, information on the macroblock in the previous frame is needed.

Basically, an existing rectangular image (MPEG1, M
The following three states can be considered when the states at the time of shape-encoding the bit stream of the PEG2 encoded image) are summarized. (1) The block of the object has been intra-coded (without reference to others).

(2) An object block is generated with reference to an image in the object.

(3) The block of the object is generated with reference to a part or the whole of the image outside the object.

Originally, since shape coding targets an image object, it is considered that the states (1) and (2) are almost the same. In this case, only the conversion of the bit stream is performed without re-coding. It is possible to create a bit stream of arbitrary shape encoding with. The problem is the case (3), which cannot produce an arbitrary shape encoded bit stream without re-encoding.

Therefore, in order to solve the case (3) efficiently, the following method is used in this embodiment. In this case, since a pixel at a position that is not shape-coded is basically referred to, the decoded macroblock is re-encoded and predicted according to a predicted frame. That is, the unidirectional frame is unidirectional prediction or intra-coding, and the bidirectional predicted frame is bidirectional prediction, unidirectional prediction, or intra-coding.

The bit stream editing section 4 of this embodiment has a function of efficiently solving the above-mentioned case (3), and the configuration is as shown in FIG. Existing bitstream 10
0 is sent to the target block information extraction unit 41 necessary for shape encoding. Target macroblock information extraction unit 41
Receives the object position information (shape signal 60)
The intra-macroblock information and the motion vector are extracted from the bit stream 100, and the motion vector is reconstructed (construction of a real vector).
Send to 2.

The vector range determination unit 42 determines whether or not the macroblock refers to an image in the macroblock constituting the object from the object position information 60, and determines the macroblock information of the existing bit stream 100. It determines whether to generate or re-encode the shape-encoded bit stream using this, and operates the switch unit 44.

The switch unit 44 switches to the terminal a when forming a shape-encoded bit stream using the macroblock information of the existing bit stream 100, and switches to the terminal b when re-encoding.

On the other hand, the target block information reconstructing unit 43
It receives the information in the macroblock, generates the skipped macroblock, and completely reconstructs the macroblock information. At this time, the target block information reconstructing unit 43 also reconstructs the motion vector information, and outputs both to the subsequent-stage shape-encoded bitstream creating unit 46 via the switch unit 44.

The macroblock information shape-encoded by the shape encoding unit 45 controlled so as to be completed in the object for an image that does not exist in the object is also sent to the switch unit 44.

Therefore, when forming a shape-encoded bit stream using the macro-block information of the existing bit stream 100, the macro-block information from the target block information reconstructing unit 43 The macroblock information re-encoded by the shape encoding unit 45 when input to the stream creation unit 46 and re-encoded is input to the shape-encoded bit stream creation unit 46 via the switch unit 44.

Shape-encoded bit stream creation unit 46
Is an existing bit stream 1 in macroblock units.
By creating a shape-encoded bit stream using either the intra-macroblock information reconstructed from 00 and the macro block newly created by the shape encoding unit 45, the bit stream 30 in object units is created.
Outputs 0.

According to the present embodiment, the existing bit stream 100 is used as much as possible, and only when it is necessary to refer to a macroblock outside the object, the temporarily decoded and re-encoded bit stream is used. By creating the per-object bitstream 300, most of the per-object bitstream can be constructed using existing non-recoded bitstreams, and without loss of image quality, Only bitstreams can be created.

[0040]

As described in detail above, according to the first aspect of the present invention, when extracting necessary objects from an existing bit stream and performing shape coding, all the existing bit streams are used. Accordingly, a bit stream of only a designated arbitrary object can be created from an existing bit stream without deteriorating an image at all.

According to the second or third aspect of the present invention, when extracting necessary objects from an existing bit stream and performing shape coding, the existing bit stream is used as much as possible, so that the image is degraded. Without letting
It is possible to create a bit stream of only the specified arbitrary object.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment of an encoding device according to the present invention.

FIG. 2 is a block diagram illustrating a detailed configuration example of a bit stream editing unit illustrated in FIG. 1;

FIG. 3 is a diagram showing a relationship between an object and a macroblock.

FIG. 4 is a diagram illustrating a relationship between an object of a previous frame and a macroblock and a relationship between an object of a current frame to be encoded and a macroblock.

FIG. 5 is a block diagram showing a conventional method for creating a bit stream of only a specific object from an existing bit stream.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Decoding part 2 Object extraction part 3 Object position information part 4 Bit stream editing part 41 Target macroblock information extraction part 42 Vector range judgment part 43 Target block information reconstruction part 44 Switch part 45 Shape coding part 46 Shape coding bit Stream creation unit

Claims (3)

[Claims] 1. An encoding apparatus for generating a bit stream that forms only an object arbitrarily designated separately from an existing encoded bit stream, comprising: An encoding apparatus comprising: means for extracting a group of blocks forming an object, and forming a bit stream shape-encoded only by the extracted group of blocks. 2. The encoding apparatus according to claim 1, wherein, when the position information of the object indicates that a macroblock constituting the object refers to an image outside the object, a block of the corresponding image is determined. An encoding apparatus, comprising: forming a shape-encoded bit stream using blocks obtained by re-encoding. 3. Decoding means for decoding an existing coded bit stream to obtain image data, and object extracting means for extracting an outline of an object specified separately from the image data obtained by the decoding means. Object position information generating means for generating a shape signal in macroblock units corresponding to the object from the contour information of the object obtained by the object extracting means, based on the shape signal obtained by the object position information generating means When it is determined that the macroblocks constituting the object are generated with reference to only the image in the object, the macroblocks constituting the object are used by using the existing bitstream. Generated with reference to the image And a bit stream editing means for creating a shape-encoded bit stream using a bit stream obtained by re-encoding the image data obtained by the decoding means. Encoding device.