JP3905938B2 - Image coding apparatus and image decoding apparatus - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an apparatus for compressing and encoding an image into a small amount of information and compression in a system for transmitting / accumulating an image signal such as a TV phone, a TV conference system, a portable information terminal, a digital video disk system, and a digital TV broadcast system. The present invention relates to an apparatus that restores encoded information and reproduces an image, and particularly has high error tolerance and high-quality image transmission even when an encoded image is transmitted / stored via an error-prone medium such as a wireless transmission path. The present invention relates to an image encoding device and a decoding device that can perform storage.
[0002]
[Prior art]
Techniques for compression coding to reduce the amount of information for transmitting and storing image signals, such as discrete cosine transform (DCT) coding, subband coding, pyramid coding, variable length coding, and combinations of these Various methods have been developed. In addition, ISO / JPEG is used for encoding still images, and ISO / MPEG1, MPEG2 and ITU-T / H. 261, H.M. 262, H.C. H.263 is defined. These are all systems using cosine transform coding, and are described in detail in Reference 1: edited by Hiroshi Yasuda, “International Standard for MPEG / Multimedia Coding”, Maruzen (September 1994), etc. ing.
[0003]
By encoding an image signal into a small amount of information using compression encoding technology, for example, it becomes possible to transmit an image on a transmission line having a limited transmission line capacity such as a wireless line, for example, a wireless transmission function. Image communication of still images and moving images is possible between a TV phone device, a personal digital assistant, a personal computer, a workstation, and the like to which is added. However, in general, there is a high probability that a transmission path error is mixed in a wireless transmission path, and when a compression-coded image is transmitted through the wireless transmission path, the decoded reproduced image may be deteriorated due to the influence of the transmission path error. is there.
[0004]
In conventional image coding methods such as the international standard method described above, the input image signal is divided into small areas called blocks or macroblocks, and prediction codes are used. Fake Alternatively, after performing motion compensation predictive coding, variable length coding is performed on coding mode information, motion vector information, DCT coefficient information, and the like to generate a compressed code, which is multiplexed and transmitted as a coded bit stream.
[0005]
In such an image coding method using variable length coding, a phenomenon called loss of synchronization of variable length decoding may occur in the decoding device due to mixing of transmission path errors. When loss of synchronization occurs, a code string not included in the code table is detected due to a transmission path error in variable-length decoding, or a state that violates predetermined syntax and semantics is detected. When such a phenomenon occurs, information cannot be decoded at all until the next synchronization code is detected, so that the reproduced image is greatly degraded.
[0006]
In the conventional image coding method, motion vector information, DC component information of DCT coefficients, and the like are encoded by taking a difference from an adjacent block or macroblock. In this case, if an error is mixed into certain motion vector information, DC component information, etc. and the decoded value becomes an incorrect value, the effect of the error also affects the information obtained by taking the difference from these information. The quality of the reproduced image deteriorates over a wide area in the screen.
[0007]
In order to suppress such a phenomenon, in the conventional image coding method, a synchronization code is added to each small region, and the range for obtaining the difference is limited to the small region. A technique called GOB / slice that suppresses the spread of image quality degradation due to the conversion to a small area is used.
[0008]
FIG. 9 shows an example of area division in GOB / slice. In the figure, a small square divided by a solid line or a broken line is a macroblock. In addition, a region surrounded by a solid line is GOB / slice, and a macroblock indicated by a dot pattern in the figure is a macroblock to which a GOB / slice synchronization code can be added. However, these conventional methods have the following problems.
[0009]
In the GOB of FIG. 9A, a synchronization code can be added only to the macro block at the left end of the screen. Therefore, the unit of the area that suppresses the spread of errors in the synchronization code is a large area such as one or a plurality of macroblock lines. For this reason, the range in which the error spreads is wide, and the image quality is greatly deteriorated.
[0010]
In the slice of FIG. 9B, a synchronization code can be added to all macroblocks. For this reason, the range in which the error spreads can be reduced, but on the other hand, a large number of bits are required to transmit the position information indicating the position to which the synchronization code is added, and the code amount increases.
[0011]
In the method of adding the synchronization code in FIG. 9C to each fixed macroblock, the number of fixed macroblocks (3 in the example in the figure) is not a divisor of the number of horizontal macroblocks in the screen (11 in the example in the figure). As shown in the figure, there is a problem that the horizontal position of the macroblock into which the synchronization code can be inserted differs depending on the vertical position of the macroblock.
[0012]
[Problems to be solved by the invention]
As described above, in a conventional image encoding / decoding apparatus that adds a synchronization code to an encoded bitstream by a method such as GOB / slice and suppresses the spread of image quality degradation to a small area, the unit to which the synchronization code can be added is The amount of code increases because a large number of bits are required for transmission of position information indicating the position to which the synchronization code is added, and the problem that the error spreads over a wide range and the image quality is greatly deteriorated. There was a problem of doing.
[0013]
An object of the present invention is to solve such a conventional problem and to provide an image encoding device and an image decoding device that can increase error resilience without greatly increasing the amount of codes.
[0014]
[Means for Solving the Invention]
A first image encoding device according to the present invention includes an image encoding unit that compresses and encodes an image signal and outputs a compressed code, and multiplexes the compression code output from the image encoding unit and outputs a synchronization code. And a multiplexing / synchronizing addition unit that outputs a coded bit stream by adding the multiplexing / synchronizing addition unit only at a plurality of specific positions in which at least horizontal positions in the screen are intermittently different. It is configured so that a synchronization code can be added, and when the synchronization code is added, position information indicating a position where the synchronization code is added among a plurality of specific positions is added to the encoded bitstream. And
[0015]
The first image decoding apparatus corresponding to the first image encoding apparatus detects a synchronization code in the input encoded bit stream and demultiplexes the compression code in the encoded bit stream. Synchronization detection / demultiplexing means, and image decoding means for decoding a demultiplexed compressed code output from the synchronization detection / demultiplexing means to reproduce an image signal, The synchronization code is added only to at least one of a plurality of specific positions in which at least the horizontal position in the screen is intermittently changed. The synchronization detection / demultiplexing means And detecting position information indicating a position to which the synchronization signal is added among a plurality of specific positions.
[0016]
In the first image encoding device / decoding device, the position where the synchronization code can exist is set as a plurality of specific positions in which at least the horizontal position in the screen is intermittently changed. Therefore, like the conventional GOB The synchronization code can be added without restriction that the synchronization code can be added only to one or a plurality of macroblock lines. Therefore, the influence of the transmission path error can be limited to a narrow range, and the quality of the reproduced image is improved. In addition, the number of macroblocks to which a synchronization code can be added is small compared to the conventional method using slices, and the information for identifying this position can be suppressed to a small number of bits, so that the amount of codes is reduced.
[0017]
The second image encoding apparatus according to the present invention includes an image encoding unit that compresses and encodes an image signal and outputs the compressed code, and multiplexes the compression codes output from the image encoding unit and adds a synchronization code. And a multiplexing / synchronization adding means for outputting a coded bitstream, and the multiplexing / synchronization adding means is configured to add a synchronization code only to a plurality of specific positions in the screen, When the first position information indicating the specific position is included in the encoded bitstream and a synchronization code is further added, the second position information indicating the position where the synchronization code is added among the plurality of specific positions is encoded. It is characterized in that it is added to the digitized bit stream.
[0018]
The first image decoding apparatus corresponding to the first image encoding apparatus detects a synchronization code in the input encoded bit stream and demultiplexes the compression code in the encoded bit stream. Synchronization detection / demultiplexing means, and image decoding means for decoding a demultiplexed compressed code output from the synchronization detection / demultiplexing means to reproduce an image signal, The synchronization code is added only to at least one of a plurality of specific positions in the screen, and the synchronization detection / demultiplexing means detects first position information indicating the specific position in the encoded bitstream. Based on this, a synchronization code is detected, and second position information indicating a position to which the synchronization signal is added among a plurality of specific positions is detected.
[0019]
In the second image encoding device / decoding device, since information indicating a specific position in the screen to which a synchronization code can be added is included in the encoded bitstream and encoded, for example, an input image to be encoded is encoded. It is possible to control the position and number of synchronization codes that can be added according to the nature of the signal, the nature of the transmission path for transmitting the encoded bit stream, and the like. For this reason, synchronization codes can be added more effectively than conventional GOBs and slices, and the quality of reproduced images is improved.
[0020]
In the present invention, a plurality of specific positions in the screen to which a synchronization code can be added may be regularly determined, or at least one of a contour portion in the screen and a predetermined specific region, specifically The position may be set in a position surrounding an object such as a person on the screen or in the object.
[0021]
By doing so, it is possible to add a synchronization code so as to more strongly protect a person or other information, which is visually important information, and to improve the quality of a reproduced image when a transmission path error occurs.
[0022]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the drawings.
[0023]
(First embodiment)
FIG. 1 is an example of a moving image high-efficiency compression coding apparatus using discrete cosine transform coding, which is a kind of motion compensation adaptive prediction and orthogonal coding. Since the motion compensation adaptive prediction discrete cosine transform coding method is detailed in the above-mentioned document 1 and the like, only the outline of the operation will be described, and the difference from the conventional method will be described in detail.
[0024]
In FIG. 1, an input moving image signal 131 to be encoded input in units of frames is first subjected to motion compensation adaptive prediction in units of small regions such as macroblocks. That is, the motion compensated adaptive predictor 101 detects a motion vector between the input moving image signal 131 and the image signal that has been stored in the frame memory 102 and has already been encoded / locally decoded. A prediction signal 132 is generated by motion compensation prediction based on the vector. In the motion compensated predictor 101, a prediction mode which is suitable for encoding among motion compensation predictive coding and intraframe coding (predicted signal = 0) in which the input moving image signal 131 is coded as it is. Is selected, and a corresponding prediction signal 132 is output.
[0025]
The prediction signal 132 is input to the subtractor 103, and the prediction residual signal 133 is output by subtracting the prediction signal 132 from the input moving image signal 131. The prediction residual signal 133 is subjected to discrete cosine transform (DCT) in units of blocks having a certain size in the discrete cosine transformer 104, and DCT coefficients are generated. This DCT coefficient is quantized by the quantizer 105. The quantized DCT coefficient data from the quantizer 105 is divided into two branches, one of which is variable length encoded by the first variable length encoder 106 and the other is dequantized by the inverse quantizer 107. Then, the inverse discrete cosine transformer 108 performs inverse discrete cosine transform (inverse DCT), thereby decoding the prediction residual signal. The prediction residual signal output from the inverse discrete cosine transformer 108 is added to the prediction signal 132 in the adder 109 to generate a local decoded signal. This locally decoded signal is stored in the frame memory 102.
[0026]
On the other hand, the prediction mode and motion vector information determined by the motion compensated adaptive predictor 101 are variable length encoded by the second variable length encoder 110. However, the difference between the motion vector information and the predicted motion vector obtained from the adjacent already encoded motion vector information is obtained, and this is variable-length encoded.
[0027]
The variable length codes (compression codes) output from the first and second variable length encoders 106 and 110 are the same as the compression codes output from the variable length encoders 106 and 110 in the multiplexing / synchronization adder 111. Multiplexing and addition of a synchronization code are performed, and an encoded bit stream 205 is obtained.
[0028]
Next, the multiplexing / synchronization adder 111 will be described.
FIG. 2 is a diagram showing a flow of multiplexing in the multiplexing / synchronizing adder 111. Multiplexing is a block, a macroblock (MB) composed of a plurality of blocks, a group of macroblocks (also referred to as GOB, slice, etc.) composed of one to a plurality of macroblocks, or a picture (screen) composed of one to a plurality of GOMs. (Also referred to as a frame). Furthermore, layers such as a GOP (Group of Pictures), a sequence, a session, or the like composed of one or a plurality of pictures may be used.
[0029]
As shown in FIG. 2A, the picture layer starts with a picture synchronization code (PSC), a picture header (PH) indicating a coding mode of the picture, and is followed by one or more GOM layers.
As shown in FIG. 2 (b), the GOM layer starts with a GOM synchronization code (GSC) and a GOM header (GH) indicating the position of the GOM in the screen, a coding mode, etc. One or more contained macroblocks follow.
As shown in FIG. 2C, the macroblock layer includes coding mode information, motion vector information, and a plurality of blocks included in the macroblock.
[0030]
FIG. 3 is a diagram illustrating an example of information included in the GOM header (GH). The GOM header indicates encoding mode information GM indicating the encoding mode of the GOM, position information GN indicating the position in the GOM picture (screen), and a quantization step size of the DCT coefficient included in the GOM. Consists of step size information GQ.
[0031]
FIG. 4 is a diagram showing two examples of GOM shapes. In the figure, a solid line indicates a GOM boundary, and a broken line indicates a macroblock boundary in the GOM. In GOM Diagonal line The first macroblock indicated by (2) is a specific position where the GOM synchronization code GSC can be added, that is, the GOM synchronization code GSC can be added. However, since the picture synchronization code PSC is added to the first GOM in the screen, the GOM synchronization code GSC is not added.
[0032]
In the example of FIG. 4A, the GOM synchronization code can be added to the 0th, 3rd, 6th, and 9th macroblocks with the left end of the position in the horizontal direction of the screen being 0. In the example of FIG. 4B, the position in the horizontal direction of the screen is 0, 2, 4, 6, 8, 10th with the left end as 0, and the position in the vertical direction of the screen is 0, 2, 4, 4, with the upper end as 0. A GOM synchronization code can be added to the sixth and eighth macroblocks. If the positions of the macroblocks in the horizontal and vertical directions are Mx and My, in the example of FIGS. 4A and 4B, a GOM synchronization code may be added to a macroblock at a position where the following conditions are satisfied. it can.
[0033]
(A) (Mx mod 3) = 0
(B) ((Mx mod 2) = 0) and ((My mod 2) = 0)
Here, (A mod B) represents a remainder when A is divided by B.
[0034]
In FIGS. 4A and 4B, a GOM synchronization code can be added. Diagonal line Are numbered in order from the upper left of the screen, and these numbers are added to the encoded bitstream 205 as the position information GN in the syntax of FIG. That is, the position information GN in this case indicates a specific position to which a GOM synchronization code can be added.
[0035]
By arranging a GOM as in the example of FIGS. 4A and 4B, a synchronization code is added only for one or several macroblock lines as in the conventional GOB shown in FIG. 9A. Compared with a method that cannot be used, the density at which a synchronization code can be added is high, and the range in which errors are spread is reduced, so that the quality of a reproduced image is improved.
[0036]
Further, in the conventional slice shown in FIG. 9B, the total number of macroblocks to which a synchronization code can be added is large, and when there are 99 macroblocks as in the example of FIG. In order to specify the position where the synchronization code is added to all, numbering indicated by the position information GN in FIG. 3 is necessary, and it is necessary to allocate a large number of bits to the position information GN.
[0037]
On the other hand, in the example of FIG. 4, the total number of macroblocks to which a synchronization code can be added is 35 in the case of FIG. 4A and 29 in the case of FIG. 4B. The number of bits can be reduced. For example, if GN is expressed in a binary representation of a fixed length, 7 bits are required for a slice, whereas 6 bits are required for the example in FIG. 4A and 5 bits are required for the example in FIG. 4B. This leads to a large code amount reduction especially when a large number of synchronization codes are added.
[0038]
Next, the moving picture decoding apparatus according to the present embodiment will be described.
[0039]
FIG. 5 is a block diagram showing a configuration of a moving picture decoding apparatus corresponding to the moving picture decoding apparatus of FIG. In FIG. 5, the encoded bit stream 205 output from the moving image encoding apparatus in FIG. 1 passes through a transmission / storage system (not shown) and then input to the synchronization detection / demultiplexer 811 as an encoded bit stream 205 ′. Is done.
[0040]
The synchronization detector / demultiplexer 811 detects a synchronization code in the encoded bitstream 205 ′, and also includes a variable length code (hereinafter referred to as a prediction residual code) 841 of a quantized DCT coefficient, a prediction mode, and A variable length code (hereinafter referred to as motion compensation adaptive prediction information code) 842 of motion vector information is separated and output.
[0041]
The prediction residual code 841 is input to the first variable length decoder 806, and the quantized DCT coefficient 831 is decoded. The DCT coefficient 831 is an output from the motion compensated adaptive predictor 801 after being subjected to a series of processing such as inverse quantization by the inverse quantizer 807 and discrete cosine transform by the inverse DCT 808. The reproduced image signal 850 is added to the motion compensation adaptive prediction signal 832. The reproduced image signal 850 is output outside the apparatus and recorded in the frame memory 820.
[0042]
On the other hand, the motion compensation adaptive prediction information code 842 is input to the second variable length decoder, and the prediction mode and motion vector information are decoded. The decoded prediction mode and motion vector information are input to the motion compensated adaptive predictor 801, and a motion compensated prediction signal 832 is generated based on the information and the image signal from the frame memory 820.
[0043]
The above processing is processing for reproducing a moving image corresponding to the moving image encoding device of FIG. 1, and the processing performed by the inverse quantizer 807, the inverse DCT device 808, the adder 809, and the frame memory 820, respectively. The implementation means of the inverse quantizer 107, inverse DCT unit 108, adder 109, and frame memory 102 in FIG. 1 may be different from each other, but are basically the same. Also, the processes of the first and second variable length decoders 806 and 810 and the synchronization detector / demultiplexer 811 are the same as the variable length shown in FIG. 1 except when an error is mixed in the encoded bit stream. This process is the reverse of that of the encoders 106 and 110 and the multiplexer / synchronizer adder 111.
[0044]
Next, the synchronization detector / demultiplexer 811 will be described.
[0045]
The synchronization detector / demultiplexer 811 performs synchronization code detection and demultiplexing processing based on the multiplexing rule shown in FIG. The GOM synchronization code GSC and the GOM header GH may or may not exist depending on the GOM. Therefore, in FIG. 4 (a) (b) Diagonal line It is determined whether or not the GOM synchronization code GSC exists at the time when the decoding of the macroblock to which the GOM synchronization code shown in FIG.
[0046]
The synchronization code is a code word called a unique word that does not have the same bit pattern as other code strings. Since a plurality of types of synchronization codes (two types of picture synchronization code PSC and GOM synchronization code GSC in FIG. 2) are different codewords, the first codeword in each GOM code string is a GOM synchronization code. Can be determined uniquely.
[0047]
When the GOM synchronization code GSC is detected, the synchronization detector / demultiplexer 811 also decodes the subsequent GOM header GH. Thereafter, the demultiplexing process of the macroblock layer included in the GOM is performed. Here, when it is found in the decoding process of the GOM header or the decoding process of the macroblock layer that a transmission path error is mixed in the encoded bitstream 205 ′, the subsequent macroblock in the GOM Starts the process of detecting the synchronization code included in the subsequent code string without performing the decoding process.
[0048]
If the detected synchronization code is the GOM synchronization code GSC, the GOM header of the GOM is decoded, and the position in the GOM screen indicated in the GOM header (the GOM synchronization code GSC is added). The position of the GOM, that is, the position to which the GOM synchronization code GSC is added is specified based on the position information GN indicating the position, and the decoding process is performed.
[0049]
If the detected synchronization code is the picture synchronization code PSC, decoding of the picture is aborted at the GOM where the error is found or the previous GOM, and the picture where the picture synchronization code is detected is decoded.
[0050]
In the present embodiment, the addition of the GOM synchronization code GSC in the multiplexing / synchronization adder 111 in FIG. 1 may be performed by all GOMs or only a part of GOMs. Alternatively, the GOM synchronization code may not be added at all. Adding more GOM synchronization codes increases the tolerance to transmission path errors, and when a transmission path error occurs with a high probability, the quality of the reproduced image in the video decoding apparatus in FIG. 5 is improved. However, if a large number of GOM synchronization codes are added, the amount of codes accompanying the GOM synchronization code and the GOM header increases, and the coding quality when there is no transmission path error decreases. It is also effective to adaptively change whether or not is added according to the quality of the transmission path, the coding rate, the nature of the input video signal, and the like.
[0051]
(Second Embodiment)
Next, a second embodiment according to the present invention will be described. The moving picture coding apparatus according to the present embodiment determines a GOM shape (a position of a macroblock into which a GOM synchronization code can be inserted) according to an input moving picture signal, a motion vector, and a prediction mode, and GOM shape information ( (Position information indicating the position of a macroblock into which a GOM synchronization code can be inserted) is included in the encoded bitstream and encoded. In addition, the moving picture decoding apparatus corresponding to this decodes GOM shape information (position information indicating the position of a macroblock into which a GOM synchronization code can be inserted) included in the input encoded bitstream, and uses this The video is decoded.
[0052]
FIG. 6 is a block diagram of the video encoding apparatus according to this embodiment. The same reference numerals are assigned to the parts corresponding to those in FIG. 1, and the differences from the first embodiment will be mainly described. In this embodiment, a GOM shape determiner 601 is newly added, and this GOM shape determination is performed. The GOM shape information 602 is output from the encoder 601 to the multiplexing / synchronous code encoder 111.
[0053]
The GOM shape determiner 601 determines the GOM shape based on the input moving image signal 131, the motion vector determined by the motion compensated adaptive predictor 101, information on the encoding mode, and the like, and outputs the GOM shape information 602. The GOM shape information 602 is variable-length encoded by the multiplexing / synchronous code encoder 111 and added to the encoded bit stream (output code string) 205.
[0054]
Next, the GOM shape determiner 601 will be described.
FIGS. 7A and 7B show an example in which object detection is performed on the input moving image signal 131 and the GOM shape is determined based on the object detection. First, object detection in the input moving image signal 131 is performed. This object detection may be performed directly from the input moving image signal 131 or may be determined based on a motion vector or a coding mode. In FIGS. 7A and 7B, the region indicated by the alternate long and short dash line is the detected object. A macro block to which a GOM synchronization code can be added is added to a boundary between objects, a boundary between an object and a background, that is, a macro block in an outline portion (in the figure). Diagonal line Macroblock). In this case, position information indicating the position of a macroblock to which a GOM synchronization code can be added is added to the encoded bitstream 205 as GOM shape information 602.
[0055]
By determining the GOM shape in this way, generally, the properties of an image are greatly different, and if an error spreads, the image quality is greatly deteriorated, and the spread of an error between the object and the background or between different objects can be suppressed. . For this reason, it is possible to prevent the reproduction image from being deteriorated in the decoding apparatus when a transmission path error occurs.
[0056]
In these cases, the density of macroblocks to which the GOM synchronization code can be added may be changed for each object. In the example of FIG. 7B, the density of macroblocks to which a GOM synchronization code can be added in a specific region, in this example, a portion corresponding to a human face is high, and more GOM synchronization codes are added. Is possible. Since the face of a person is subjectively very important information, the subjective quality is improved by adding more GOM synchronization codes and increasing the error tolerance of the portion. On the other hand, in a less important part such as the background, the density of macro blocks to which the GOM synchronization code can be added is lowered, the number of macro blocks to which the GOM synchronization code of the entire screen can be added is reduced, and the code amount of the GOM header is reduced. Can be reduced. As the candidates, the GOM shape shown in FIG. 4 or the GOM shape corresponding to the conventional GOB / slice shown in FIG. 9 may be included.
[0057]
The macroblock to which the GOM synchronization code in the GOM shape determiner 601 can be added may be selected and determined from a plurality of candidates prepared in advance. In this case, information related to this selection decision is encoded as GOM shape information 602.
[0058]
The GOM shape information may be encoded by being included in the picture header PH. If there is a higher layer (GOP layer, session layer, sequence layer, etc.) than the picture, it is included and encoded Also good.
[0059]
Next, the moving picture decoding apparatus according to the present embodiment will be described.
FIG. 8 is a block diagram showing a configuration of a moving picture decoding apparatus corresponding to the moving picture encoding apparatus of FIG. When the same reference numerals are assigned to the parts corresponding to those in FIG. 5 and the differences are mainly described, a GOM shape decoder 815 for decoding GOM shape information is newly added in the present embodiment.
[0060]
The synchronization detector / demultiplexer 811 demultiplexes the GOM shape information code 843 obtained by variable-length coding the GOM shape information included in the input encoded bitstream 205 ′, and outputs the demultiplexed data to the GOM shape decoder 815. To do. The GOM shape decoder 815 decodes the GOM shape information code 843 and outputs the GOM shape information 839 to the synchronization detection / demultiplexer 811.
[0061]
The synchronization detector / demultiplexer 811 detects the GOM synchronization code and decodes the GOM header based on the decoded GOM shape information 839.
[0062]
Since the association between the position on the GOM screen and the GN in FIG. 3 is determined by the GOM shape, the moving picture decoding apparatus according to the present embodiment performs decoding based on the decoded GOM shape information 839 and the GN value. Specify the position in the GOM screen. Furthermore, since the macroblock to which the GOM synchronization code can be added is specified by the decoded GOM shape information, it is also possible to determine whether or not the GOM synchronization code as described in the first embodiment has been added. This macro block can be used.
[0063]
In the present embodiment, the GOM shape may be determined according to the code amount. For example, the GOM shape is determined so that the code amount of each GOM is substantially constant. Specifically, the GOM shape determiner 601 counts the sum of the code amounts of the macroblocks in the GOM, and the sum is a predetermined value, or the characteristics of the input video signal, the transmission path, etc. When the value determined in accordance with the above is exceeded, the macro block may be determined as the last macro block in the GOM.
[0064]
In the image coding method in which the prediction value is obtained from the information in the adjacent macroblock when coding the information of the motion vector information and the DC component of the DCT coefficient, and coding is performed by taking the difference between them, different GOMs are used. The motion vector information, DC component information of the DCT coefficient, and the like may be used only for data in the same GOM without using the prediction value. As a result, even if a transmission path error is mixed in other GOM data, the influence is not spread, so that the quality of a reproduced image when there is a transmission path error is improved. In the video encoding apparatus according to the present invention, the GOM shape can be controlled according to the nature of the input video signal, motion vector information, encoding mode information, DCT coefficient components, and the like. Even if it goes, the fall of prediction efficiency can be suppressed small.
[0065]
That is, the GOM shape determiner 601 may determine the GOM shape so that the increase amount of the code amount when the prediction is limited to the code amount when the limitation is not performed is small. For this, control may be performed based on the amount of code when encoding is actually performed, or control is performed so that the size of the prediction residual is not as large as compared with the case where no restriction is made. Also good.
[0066]
In an encoding method for obtaining a prediction value from data of a macroblock on the left side or a plurality of macroblocks including the left side, in the macroblock on the left side of the screen, the macroblock on the left side does not exist. Prediction efficiency decreases compared to macroblocks. Therefore, by increasing the macro block at the left end of the screen to the first macro block of GOM, that is, the macro block into which the synchronization code can be inserted, the increase in the code amount due to the intra-GOM restriction of prediction can be reduced.
[0067]
The GOM shape control in consideration of these prediction limitations may be used in combination with the GOM shape control in consideration of the code amount described above. For example, the GOM shape determiner 601 counts the total code amount of each macroblock in the GOM, and the sum is determined in accordance with a predetermined value or the characteristics of the input moving image signal or the transmission path. If the specified value is exceeded, the macro block is determined as the last macro block candidate in the GOM. Next, this candidate macroblock and the macroblock at the left end of the screen to be encoded before and after are selected, and this is determined as the first macroblock of the next GOM, that is, the macroblock into which the synchronization code can be inserted. May be.
[0068]
【The invention's effect】
As described above, according to the present invention, the position at which the synchronization code can be added is set to a plurality of specific positions in which at least the horizontal position in the screen is intermittently changed. In addition, a synchronization code can be added without being restricted by the fact that a synchronization code can be added only for one or a plurality of macroblock lines. Therefore, the influence of transmission path errors can be limited to a narrow range, the quality of the reproduced image is improved, and the number of macroblocks to which a synchronization code can be added is reduced compared to the conventional method using slices. Since the information for identifying the number can be suppressed to a small number of bits, the amount of codes can be reduced.
[0069]
In addition, according to the present invention, information indicating a specific position in the screen to which a synchronization code can be added is included in the encoded bitstream and encoded, for example, the nature of the input image signal to be encoded, Since it is possible to control the position and number of synchronization codes that can be added according to the nature of the transmission path for transmitting the digitized bit stream, the synchronization code is more effective than conventional GOBs and slices. Can be added, and the quality of the reproduced image is improved.
[0070]
In addition, a plurality of specific positions in the screen to which a synchronization code can be added are defined in a contour portion or a predetermined specific area in the screen, that is, a position surrounding an object such as a person in the screen, or in the object. By defining, it is possible to add a synchronization code so as to more strongly protect people and other information, which is visually important information, and to improve the quality of a reproduced image when a transmission path error occurs. it can.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a configuration of a moving image encoding apparatus according to a first embodiment of the present invention.
FIG. 2 is a diagram showing a multiplexing rule in a multiplexing / synchronous adder in the moving picture encoding apparatus of FIG. 1;
3 is a diagram showing information included in a GOM header of an encoded bit stream in the moving image encoding apparatus in FIG. 1. FIG.
4 is a diagram showing an example of a macro block to which a synchronization code can be added in the moving picture encoding apparatus of FIG. 1;
FIG. 5 is a block diagram showing a configuration of a moving picture decoding apparatus according to the first embodiment of the present invention.
FIG. 6 is a block diagram showing the configuration of a video encoding apparatus according to the second embodiment of the present invention.
7 is a diagram showing an example of a macro block to which a synchronization code can be added in the moving picture encoding apparatus of FIG. 6;
FIG. 8 is a block diagram showing a configuration of a moving picture decoding apparatus according to the second embodiment of the present invention.
FIG. 9 is a diagram illustrating a macroblock to which a synchronization code can be added in a conventional moving image encoding device and decoding device;
[Explanation of symbols]
101 ... Motion compensated adaptive predictor
102: Frame memory
104: Discrete cosine transformer
105 ... Quantizer
107: Inverse quantizer
108: Inverse discrete cosine transformer
111 ... Multiplexing / synchronous adder
131 ... Input video signal
205, 205 '... coded bit stream
601 ... GOM shape determiner
602 ... GOM shape information
801 ... Motion compensation adaptive predictor
807 ... Inverse quantizer
808 ... Inverse discrete cosine transformer
811: Synchronization detector / demultiplexer
820: Frame memory
850 ... reproduced image signal
815 ... GOM shape information decoder
843 ... GOM shape information code
839 ... GOM shape information

Claims (4)

Image encoding means for compressing and encoding an image signal and outputting the compressed code;
Multiplexing / synchronizing addition means for multiplexing the compressed code output from the image encoding means and adding a synchronization code to output an encoded bit stream;
The multiplexing / synchronization adding means is configured to allow the synchronization code to be added only to a plurality of specific macroblocks in a screen, and includes first position information indicating the position of the specific macroblock as the code. Second position information indicating the position of a macroblock to which the synchronization code is added in the plurality of specific macroblocks when it is included in a picture header or a layer higher than the picture in the coded bitstream and a synchronization code is further added Is added to the coded bitstream. The image coding apparatus according to claim 1, wherein the position of the specific macroblock is determined in at least one of a contour portion in a screen and a predetermined specific region. Synchronization detection / demultiplexing means for detecting a synchronization code in an input encoded bitstream and demultiplexing a compression code in the encoded bitstream;
Image decoding means for reproducing the image signal by decoding the demultiplexed compressed code output from the synchronization detection / demultiplexing means,
The synchronization code in the encoded bitstream is added only to at least one macroblock among a plurality of specific macroblocks that allow the addition of the synchronization signal in the screen,
The synchronization detection / demultiplexing means includes first position information included in a layer higher than the picture header or picture in the encoded bitstream to indicate the position of the specific macroblock in the encoded bitstream. detects, and detects the synchronization code based on the first position information the detected, the second position information indicating the position of the macroblock synchronizing signal is added in said plurality of specific positions An image decoding apparatus that detects and identifies the position of a macroblock to which the synchronization signal is added based on the detected second position information . 4. The image decoding apparatus according to claim 3, wherein the position of the specific macroblock is determined in at least one of a contour portion in a screen and a predetermined specific area.

Images