JP4819024B2 - Image encoding / decoding method and system - Google Patents

Description

The present invention relates to an image coding and decoding method and system, in particular, the encoding of the multi-view image, i.e., the system needs to be coded independently of each other and the loss of such wireless and Internet occurs It relates to an image coding and decoding methods and systems for encoding and decoding of the network communication system.

  In recent years, there has been a shift from analog signal systems to digital signal systems, and the demand for digital images is increasing. However, since a digital image has a huge amount of data as it is, an encoding technique for efficiently compressing the image is important, and an international standard algorithm is widely used.

  As an example, MPEG-2 ("Information technology-Generic coding of movie pictures and associated audio information. Video," ISO / IEC 13818-2, May 1996.) is used for digital TV. The standard includes H.264. H.264 ("Draft ITU-T Recommendation and Final Draft International Standard of Joint Video Specification (IUT-T Rec. H.264 -ISO / IEC 14496-10 AVC)," Joint Video Team of ISO / IEC MPEG & ITU-T, 2003) will be used.

  The specific compression algorithm of these image compression algorithms is such that motion compensation (MC) is first performed on the encoder side, temporal signal redundancy is eliminated, and then discrete cosine transform (DCT) or the like is performed. Entropy coding is performed after the redundancy of the spatial signal is eliminated by the frequency conversion. The decoder side performs decoding by performing the reverse process.

  However, in recent years, the standardized moving image compression algorithm has a high compression efficiency, but the amount of computation has become enormous. For example, H.M. H.264 requires a large amount of arithmetic processing, and it is known that the amount of processing arithmetic increases about 5 to 10 times compared to MPEG-2. This large amount of arithmetic processing is necessary for complicated inter-frame prediction, variable macroblock size optimization, and the like, and is necessary on the encoder side. On the other hand, even if the arithmetic processing on the decoder side increases, there is a need for applications such as sensor camera and mobile phone video processing that less processing on the encoder side is preferable. Then, Distributed Video Coding (DVC) is an encoding method that satisfies this requirement and shifts the encoder side high-load calculation processing to the encoder side.

  In addition, as shown in FIG. 7, Distributed Source Coding (DSC) encodes a plurality of correlated information sources in a distributed manner without observing each other. This is a decoding system. DSC was originally established by the Slepian-Wolf theorem (D. Slepian and JK Wolf, "Noiseless coding of correlated information sources," IEEE Trans. Inform. Theory, vol. 19, no. 4, pp.471-480, 1973) and the Wyner-Ziv theorem established by Wyner and Ziv (A. Wyner and J, Ziv, "The rate-distortion function for source coding with side information at the decoder," IEEE Trans, Inform. Theory, vol.22, no.1, pp1-10 1976), and extends these theorems to actual systems. The Slepian-Wolf theorem gives an allowable compression rate region that can be decoded without distortion when two information sources are distributed coded. The Wyner-Ziv theorem A rate distortion region is given when distortion occurs. These theorems illuminate that the compression limit for distributed encoding is equal to the case of non-distributed, although within the limits of the Slepian-Wolf and Wyner-Ziv theorems. .

  In recent years, using the Turbo code, the Slepian-Wolf theorem (J. Bajcsy and P..Mitran, "Coding for the Slepian-Wolf problem with turbo codes," in Proc. IEEE Gloval communications Conf., Vol.2, 2001, pp.1400-1404) and Wyner-Ziv theorem (A-Aaron and B. Girod, "Comprossion with oide information using turbo codes," in Proc. IEEE Data Compression Conf., 2002, pp.252-261) Research has been done on how close the limit is.

Although there are various methods for actually configuring DVC, as shown in FIG. 8, a method using a turbo code for encoding and decoding a Wyner-Ziv frame has been proposed (see, for example, Non-Patent Document 1). Conventionally, in this encoding method represented by MPEG, the prediction error signal is compressed / decompressed by using an information source encoder. However, in DVC, the prediction error signal is regarded as a certain type of error. Compress / decompress the image by protecting the image with a device and correcting the error. Due to this structure, in the DVC system, high-load inter-frame prediction processing can be transferred to the decoder side, and low-load encoding is realized.
B. Girod, A. Aaron, S. Rane and D. Rebollo-Monedero, "Distributed video coding," Proceedings of the IEEE, Pecial Issue on Video Coding and Delivery, vol. 93, no. 1, pp.71-83 , January 2005.

  As described above, DVC is expected to be effective as multi-view video coding and coding for a best-effort communication system for an IP network or the like. However, DVC has poorer encoding efficiency than MPEG and the like. It has been reported that the load becomes even higher than the encoding processing of H.264 (Z. M. Belkoura and T. Sikora, “Towards rate-decoder complexity optimization in Turbo-coder based distributed video coding,” PCS2000 Apr, 2006). In particular, the latter problem is unavoidable because the channel encoder uses a message propagation algorithm, and in order to put DVC into practical use, it is necessary to speed up the decoding process.

The present invention has been made in view of the above points, and is an image encoding / decoding method capable of high-speed processing by processing code processing in parallel while maintaining high encoding efficiency with a high-load DVC, and an object of the present invention is to provide a system.

  FIG. 1 is a principle configuration diagram of the present invention.

The present invention (Claim 1) is an image encoding / decoding system comprising an encoder side for encoding a moving image frame from an information source using DVC and a decoder side for decoding the encoded frame. ,
The encoder side
When a Wyner-Ziv frame that is compressed without using the information of the preceding and following frames and is decompressed using the preceding and following frames is input from the information source, the quantization unit 11 that divides into bit planes;
Encoding means 12 for generating a syndrome bit for a bit plane and transmitting it in parallel to the decoder side together with the bit plane;
Have
The decoder side
When a Key frame that is compressed and decompressed without using information of previous and subsequent frames is input from an information source, auxiliary information in units of blocks is generated using motion compensation in order to decode the Wyner-Ziv frame. Motion compensation means 15 for obtaining bit likelihoods suitable for parallelization from auxiliary information;
Decoding means 13 for decoding the bit plane using the syndrome bits received from the encoding means on the encoder side and the bit likelihood;
Reconstructing means 14 for reconstructing and outputting the bit plane decoded by the decoding means using auxiliary information.

The present invention (Claim 2) is an image encoding / decoding system comprising an encoder side for encoding a moving image frame from an information source using DVC and a decoder side for decoding the encoded frame. ,
The encoder side
A first encoding means for compressing the Key frame and transmitting it to the decoder side when a Key frame to be compressed and decompressed without using the information of the preceding and following frames is input from the information source;
When a Wyner-Ziv frame that is compressed without using the information of the previous and subsequent frames and expanded using the previous and subsequent frames is input from the information source, it is divided into bit planes and an error correction code is used for each bit plane. Generating a syndrome bit and transmitting only the syndrome bit in parallel to the decoder side, and
The decoder side
First decoding means for decompressing the compressed Key frame received from the first encoding means, dividing it into two paths, and outputting one as a decoded image of the Key frame;
When the path of the other divided Key frame input from the first decoding means is input , a motion vector is created from the Wyner-Ziv frame in units of blocks, and the processing is performed according to the range of the image. Motion compensation means for creating auxiliary information for decoding the Wyner-Ziv frame ;
When auxiliary information is input from the motion compensation means, bit likelihood estimation means for obtaining the bit likelihood of each bit plane;
Second decoding means for decoding the syndrome bits received from the second encoding means using a message probability propagation algorithm using the bit likelihood to obtain a quantized Wyner-Ziv frame;
Reconstructing means for reconstructing and outputting quantized Wyner-Ziv frames by performing inverse quantization using the compensation information generated by the motion compensation means.

The present invention (Claim 3) is an image encoding / decoding system comprising an encoder side for encoding a moving image frame from an information source using DVC and a decoder side for decoding the encoded frame. ,
The encoder side
A first encoding means for compressing the Key frame and transmitting it to the decoder side when a Key frame to be compressed and decompressed without using the information of the preceding and following frames is input from the information source;
When a Wyner-Ziv frame that is compressed without using the information of the previous and subsequent frames and is decompressed using the previous and subsequent frames is input from the information source, binary conversion is performed, and the bit plane is divided. A second encoding means for generating syndrome bits using an error correction code and transmitting them in parallel to the decoder side;
First decoding means for decompressing the compressed Key frame received from the first encoding means, dividing it into two paths, and outputting one as a decoded image of the Key frame;
When the other path of the first decoding means is input , a motion vector is generated in units of blocks from the Wyner-Ziv frame , and processing for matching the range of the image is performed to decode the Wyner-Ziv frame. Motion compensation means for generating auxiliary information for generating auxiliary information;
Bit likelihood estimation means for binary-converting the auxiliary information and obtaining the bit likelihood of each bit plane;
A second decoding means for decoding the syndrome bits of the Wyner-Ziv frame transmitted from the second encoding means using a message probability propagation algorithm using the bit likelihood and obtaining a quantized Wyner-Ziv frame;
A reconstruction unit that performs inverse binary conversion on the quantized Wyner-Ziv frame, performs inverse quantization using auxiliary information generated by the motion compensation unit, and outputs the result.

The present invention (Claim 4) is an image encoding / decoding system comprising an encoder side for encoding a moving image frame from an information source using DVC, and a decoder side for decoding the encoded frame. ,
The encoder side
A first encoding means for compressing the Key frame and transmitting it to the decoder side when a Key frame to be compressed and decompressed without using the information of the preceding and following frames is input from the information source;
When a Wyner-Ziv frame that is compressed without using the information of the preceding and following frames and is decompressed using the preceding and following frames is input from the information source, rate control means for obtaining the bit amount necessary for decoding,
Second encoding means for dividing the Wyner-Ziv frame into bit planes, creating syndrome bits, and transmitting in parallel to the decoder side according to the bit amount required for decoding determined by the rate control means;
First decoding means for decompressing the compressed Key frame received from the first encoding means, dividing it into two paths, and outputting one as a decoded image of the Key frame;
When the other path of the first decoding means is input , a motion vector is generated in units of blocks from the Wyner-Ziv frame , and processing for matching the range of the image is performed to decode the Wyner-Ziv frame. Motion compensation means for generating auxiliary information;
Bit likelihood estimation means for obtaining the bit likelihood of each bit plane using auxiliary information;
A second decoding means for decoding the syndrome bits of the Wyner-Ziv frame transmitted from the second encoding means using a message probability propagation algorithm using the bit likelihood and obtaining a quantized Wyner-Ziv frame;
Reconstructing means for performing inverse quantization on the quantized Wyner-Ziv frame using auxiliary information generated by the motion compensation means and outputting the result.

The present invention (Claim 5) provides the second encoding means,
Generating a generation matrix for generating syndrome bits using the bit amount determined in the rate control step, generating syndrome bits using the generation matrix, and transmitting the second generation means to the second decoding means;
In the second decoding means,
Means for generating a check matrix on the basis of the number of syndrome bits transmitted from the second encoding means or the estimated number of bits and decoding the syndrome bits using the check matrix ;

According to the present invention (Claim 6), an image code in an image encoding / decoding system including an encoder side for encoding a moving image frame from an information source using DVC and a decoder side for decoding the encoded frame is provided. A decoding / decoding method,
On the encoder side
When the Wyner-Ziv frame compressed by the quantization means without using the information of the preceding and following frames and expanded using the preceding and following frames is input from the information source, it is divided into bit planes,
The encoding means generates a syndrome bit for the bit plane, and transmits it in parallel with the bit plane to the decoder side,
On the decoder side,
When a key frame that is compressed and decompressed without using the information of the previous and subsequent frames by the motion compensation means is input from the information source, auxiliary information in block units is used using motion compensation to decode the Wyner-Ziv frame. And obtaining a bit likelihood suitable for parallelization from the auxiliary information,
The decoding means decodes the bit plane using the syndrome bits received from the encoding means on the encoder side and the bit likelihood,
The reconstructing means reconstructs the decoded bit plane using auxiliary information and outputs it.

According to the present invention (Claim 7), an image code in an image encoding / decoding system including an encoder side for encoding a moving image frame from an information source using DVC and a decoder side for decoding the encoded frame is provided. A decoding / decoding method,
On the encoder side
When a key frame that is compressed and decompressed without using the information of the preceding and following frames is input from the information source by the first encoding means, the key frame is compressed and transmitted to the decoder side,
When the Wyner-Ziv frame compressed by the second encoding means without using the information of the preceding and following frames and expanded using the preceding and following frames is input from the information source, it is divided into bit planes, Generate syndrome bits using an error correction code for each plane, send only the syndrome bits in parallel to the decoder side,
On the decoder side,
The first decoding means decompresses the compressed Key frame received from the first encoding means, divides it into two paths, and outputs one as a decoded image of the Key frame,
When the motion compensator receives the path of the other key frame divided from the first decoder, the motion compensator creates a motion vector in units of blocks from the Wyner-Ziv frame , and matches the range of the image To generate auxiliary information for decoding the Wyner-Ziv frame,
When the auxiliary information is input from the motion compensation unit by the bit likelihood estimation unit, the bit likelihood of each bit plane is obtained,
The second decoding means decodes the syndrome bits received from the second encoding means using the message likelihood propagation algorithm using the bit likelihood, and obtains a quantized Wyner-Ziv frame;
The reconstruction unit reconstructs the quantized Wyner-Ziv frame by performing inverse quantization using the compensation information generated by the motion compensation unit, and outputs the frame.

The present invention (Claim 8) provides an image code in an image encoding / decoding system comprising an encoder side that encodes a moving image frame from an information source using DVC and a decoder side that decodes the encoded frame. A decoding / decoding method,
On the encoder side
When a key frame that is compressed and decompressed without using the information of the preceding and following frames is input from the information source by the first encoding means, the key frame is compressed and transmitted to the decoder side,
When the Wyner-Ziv frame compressed by the second encoding means without using the information of the previous and subsequent frames and expanded using the previous and subsequent frames is input from the information source, the binary conversion is performed, and the bit plane To generate a syndrome bit using an error correction code for each bit plane and transmit it in parallel to the decoder side,
The first decoding unit decompresses the compressed key frame received from the first encoding unit, divides the compressed key frame into two paths, outputs one as a decoded image of the key frame, and the other as a motion compensation unit. Output to
Motion compensation means and the path of the other Key frame divided input from the first decoding means are inputted, the motion to create a vector from W yner-Ziv frames into blocks, match the range of the image Auxiliary information for decoding the Wyner-Ziv frame is generated by performing processing ,
By the bit likelihood estimation means, the auxiliary information is binary-transformed to obtain the bit likelihood of each bit plane,
The second decoding means decodes the syndrome bits of the Wyner-Ziv frame transmitted from the second encoding means by the message probability propagation algorithm using the bit likelihood, and obtains a quantized Wyner-Ziv frame And
The reconstruction unit performs inverse binary conversion on the quantized Wyner-Ziv frame, performs inverse quantization using the auxiliary information generated by the motion compensation unit, and outputs the result.

The present invention (Claim 9) provides an image code in an image encoding / decoding system including an encoder side that encodes a moving image frame from an information source using DVC and a decoder side that decodes the encoded frame. A decoding / decoding method,
On the encoder side
When a key frame that is compressed and expanded without using the information of the previous and subsequent frames by the first encoding means is input from the information source, the first encoding is performed by compressing the key frame and transmitting it to the decoder side . Steps ,
The rate controlling means is compressed without using the information of the previous and subsequent frames, the Wyner-Ziv frames are extended using preceding and following frames is input from an information source, determined Ru rate control bit amount necessary for decoding Steps ,
A second encoding step that divides the Wyner-Ziv frame into bit planes by second encoding means, creates syndrome bits, and transmits them in parallel to the decoder side according to the amount of bits necessary for decoding;
On the decoding side,
The first decoding unit decompresses the compressed key frame received from the first encoding unit, divides it into two paths, outputs one as a decoded image of the key frame, and moves the other path. A first decoding step of outputting to the compensation means;
When the other path is input from the first decoding unit by the motion compensation unit, a motion vector is generated in units of blocks from the Wyner-Ziv frame, and the Wyner-Ziv frame is processed according to the range of the image. A motion compensation step for generating auxiliary information for decoding
The bit likelihood estimation means, using the auxiliary information, the bit likelihood estimation step asking you to bit likelihood of each bit plane,
The second decoding means decodes the syndrome bits of the Wyner-Ziv frame transmitted from the second encoding means by the message probability propagation algorithm using the bit likelihood, and obtains a quantized Wyner-Ziv frame A second decoding step ,
A reconstruction step of performing inverse quantization on the quantized Wyner-Ziv frame by using the auxiliary information generated by the motion compensation unit and outputting the result is performed by the reconstruction unit .

In the second encoding step, the present invention (claim 10) generates a generation matrix for generating syndrome bits using the bit amount obtained in the rate control step, and uses the generation matrix. Generating syndrome bits and sending them to the second decoding means;
In the second decoding step, a check matrix is generated based on the number of syndrome bits transmitted from the second encoding means or the estimated number of bits, and the syndrome bits are decoded using the check matrix .

  As described above, the conventional DVC has a problem that the decoding process is very heavy. In conventional non-parallel DVC, encoding / decoding is sequentially performed for each bit plane. Since the upper bit plane information that has already been decoded at the time of decoding is used for decoding the lower bit plane, when parallel, the bit plane is parallelized. Therefore, there is a problem that the encoding efficiency is deteriorated because the information of the upper bit plane cannot be used.

  In order to solve this, in the present invention, by transmitting the bit plane in parallel from the encoding side to the decoding side, the decoding processing time is shortened, and parallelization is performed by suppressing the decrease in encoding efficiency. , Distributed Video Coding (DVC) high-load decoding processing can be accelerated. In addition, the number of bits required for decoding can be estimated on the encoder side, and syndrome bits can be sent to the decoding side in parallel based on the bit amount, thereby reducing the number of buffers and reducing the number of feedback controls. Is possible.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  First, the outline of the present invention will be described.

  FIG. 2 shows the configuration of a Wyner-Ziv encoding system that is the basis of the present invention.

  The coding system shown in the figure includes a channel coding unit 1, a channel decoding unit 2, a reconstruction unit 3, a bit likelihood estimation unit 4, an auxiliary information creation unit 5, an information source coding unit 6, and an information source decoding. It consists of part 7.

  There are various types of information source encoding unit 7, information source decoding unit 6, channel encoding unit 1, and channel decoding unit 2 in the figure, but they are often used when discussed in papers and the like. FIG. 3 shows a basic configuration (a spatial domain Wyner-Ziv encoding system) in which DVC is applied to moving image compression.

  The Wyner-Ziv encoding system shown in FIG. 3 includes a scalar quantizer 11, an RA-LDPC (Rate-Adaptive Low Density Parity Check) encoder 12, an RA-LDPC decoder 13, a reconstructor 14, and a motion compensator. 15 is composed. As shown in FIG. 3, a case where a Key frame is given as auxiliary information for decoding a Wyner-Ziv frame that determines the performance of a DVC system is considered in a paper or the like. This is because the above-mentioned document “A. Wyner and J, Ziv,“ The rate-distortion function for source coding with side information at the decoder, ”IEEE Trans, Inform. Theory, vol.22, no.1, pp1-10. It is the same as the model shown in “1976”.

In the following, we will use Wyner-Ziv in the spatial domain using FIG.
The configuration of the encoder will be described as an example. In the DVC system, the video frame is divided into a Key frame and a Wyner-Ziv frame. Here, according to FIG. 3, the Key frame is assigned an odd frame X _{2i ± 1,} and even frame X _2i is assigned to the Wyner-Ziv frames. The Key frame plays a role corresponding to an I picture used in MPEG, and is compressed / expanded without using information of previous and subsequent frames.

  On the other hand, the Wyner-Ziv frame plays a role corresponding to a B picture used in MPEG. The Wyner-Ziv frame is compressed without using the information of the previous and subsequent frames, but has the feature of being expanded using the previous and next frames. Details of the Wyner-Ziv frame compression / decompression process are shown below.

The Wyner-Ziv frame X _2i is divided into bit planes _L by the 2 ^M scalar quantizer 11. Each bit plane is converted from the most significant bit (MSB) to the RA-LDPC encoder 12 ((LDPC code: RG Gallager, "Low density parity check codes," in Research Monograph series. Cambridge, MIT Press, 1963). And a sufficient syndrome bit _{CL, i} is generated for each bit plane. In the conventional method (Non-Patent Document 1 described above), each bit plane is stored in the buffer in order, and is decoded sequentially from the upper bit plane in response to a request from the decoder side.

  In contrast, in the present invention, each bit plane is transmitted in parallel from the RA-LDPC encoder 12 on the encoder side together with the syndrome bits to the RALDP decoder 13 on the decoder side. In the present invention, the motion compensator 15 estimates in advance the amount of syndrome bits required on the decoder side.

On the other hand, in the motion compensator 15 on the decoder side, block unit compensation information (Side Information) Y ^b _2i is created from the Key frame X _{2i ± 1} by the following equation.

但し、上記の式の添え字ｂはブロック単位に処理しているそれぞれを表し、ＭＶはブロック単位のＫｅｙフレームから作成された動きベクトルであり、range[]は画像のレンジに合わせる処理を表している（例えば、画像が６ビット信号であるならば、補助情報も８ビットに収める事を表しており、この場合０〜２５５に収める）。

However, the subscript b in the above formula represents each processed in units of blocks, MV represents a motion vector created from the key frame in units of blocks, and range [] represents processing for matching the range of the image. (For example, if the image is a 6-bit signal, the auxiliary information is also stored in 8 bits. In this case, it is stored in 0-255).

Note that the method of creating auxiliary information is outside the scope of the present invention, and details of the processing are described in, for example, the document “J. Ascenso. C. Brites and F. Peraira,“ Improving Frame Interpolation with Spatial Motion Smoothing for Pixel. ^{Domain Distributed Video Coding ". 5 th} EURASIP, Slovak Republic, see the July 2005", and the like.

Next, the bit likelihood suitable for parallelization is estimated from the auxiliary information created by the motion compensator 15, and the bit likelihood is given to the RA-LDPC decoder 13. The bit likelihood is generally estimated using the following equation because the difference signal between the Wyner-Ziv frame X _2i and the auxiliary information Y _2i , that is, the prediction error signal is approximated by a Laplace distribution.

但し、αはラプラス分布のパラメータである。

Where α is a parameter of the Laplace distribution.

The RA-LDPC decoder 13 decodes the bit plane q ′ _2i quantized using the bit likelihood acquired from the motion compensator 15 and the syndrome bits C _{L, i} received from the rate adaptive LDPC encoder 12. .

Finally, the bit plane decoded by the RA-LDPC decoder 13 is sent to the reconstructor 14 and reconstructed by the following equation using the auxiliary information Y _2i input from the motion compensator 15.

このようにして、本発明では、デコード処理時間を並列処理することで短縮し、また、それに適したビット尤度を推定することで、並列化に夜符号化効率の劣化を抑えている。すなわち、本発明では、デコード時間を短縮した高効率な並列化ＤＶＣが可能となる。

In this way, in the present invention, the decoding processing time is shortened by performing parallel processing, and the bit likelihood suitable for the decoding processing time is estimated, thereby suppressing deterioration in night coding efficiency due to parallelization. That is, according to the present invention, highly efficient parallel DVC with reduced decoding time is possible.

  Further, according to the present invention, what is conventionally stored in the buffer and requires a bit request from the decoder side is predicted by predicting the syndrome bit amount required for decoding in parallel DVC with high accuracy in advance. And feedback can be reduced.

  Hereinafter, specific examples will be described with reference to the drawings.

[First embodiment]
FIG. 4 shows the basic configuration of the parallel DVC system according to the first embodiment of the present invention.

  In the figure, a plurality of data sequences (Key frame and Wyner-Ziv frame) are encoded on the encoder side, and a plurality of data sequences are received on the decoder side, and auxiliary information for decoding the Wyner-Ziv frame using the Key frame is shown. The system which has the function to create is shown.

  In the figure, the parallel DVC system decodes a Wyner-Ziv frame from a key frame, an encoder 20 having a communication encoder, an encoder 21 having an information source encoder, a decoder 22 having an information source decoder, and a key frame. Motion compensation unit 23 that performs auxiliary information necessary for the motion compensation, bit likelihood estimation unit 24 that estimates the bit likelihood from the auxiliary information, decoding unit 25 having a channel decoder, and inverse quantum using the auxiliary information It is comprised from the reconfiguration | reconstruction part 26 which converts.

  Each component will be described below.

  The information source is divided into a Key frame and a Wyner-Ziv frame, and the Key frame is input to the encoding unit 20 and compressed. The Wyner-Ziv frame is input to the encoding unit 21 and compressed. In the encoding unit 21 to which the Key frame is input, compression is performed according to an international standard such as JPEG2000, and the compressed data is sent to the decoding unit 22. On the other hand, in the encoding unit 20 to which the Wyner-Ziv frame is input, an error correction code is used for each bit plane. Then, only the generated syndrome bits are sent to the decoding unit 25 simultaneously (in parallel).

  In the decoding unit 22 to which the Key frame is input, it is decompressed according to an international standard such as JPEG2000 and divided into two paths. In one pass, it is output as a decoded image of a Key frame, and the other is sent to the motion compensation unit 23. In the motion compensation unit 23, auxiliary information is created based on, for example, the aforementioned equation (1) so that the information is as equal as possible to the Wyner-Ziv frame.

  The created auxiliary information is sent to the bit likelihood estimation unit 24 and the reconstruction unit 26. The bit likelihood estimator 24 estimates the bit likelihood of each bit plane using the above equation (2) by the following equation.

ただし、Ｚ_Ｌはあるビットが１を取り得る範囲を表しており、次式で与えられる。

However, Z _L represents a range in which a certain bit can take 1, and is given by the following equation.

上記の式（４）及び式（５）より、本システムでは、上位ビットプレーンＸ_2i ^j-1を用いないでビット尤度を推定しているために、並列処理を可能にしている。

From the above equations (4) and (5), in this system, since the bit likelihood is estimated without using the upper bit plane X _2ij ⁻¹ , parallel processing is enabled.

  In the bit likelihood estimation unit 24, using the bit likelihood derived by the equation (4), the decoding unit 25 decodes the syndrome bits created by the encoding unit 20 using a message probability propagation algorithm, Get the quantized Wyner-Ziv frame.

  Finally, reconstruction is performed by the reconstruction unit 26 using the auxiliary information and the quantized Wyner-Ziv frame as close as possible to the original Wyner-Ziv frame, and the decoded Wyner-Ziv frame is output. The This process is performed according to, for example, the equation (3).

  Through the above operation, parallel decoding processing of Wyner-Ziv frames can be performed, and high-speed decoding processing of a heavy message probability propagation algorithm can be expected.

[Second Embodiment]
FIG. 5 shows a configuration of a parallel DVC system according to the second embodiment of the present invention.

  The parallel DVC system shown in the figure encodes a plurality of data sequences (Key frame and Wyner-Ziv frame) on the encoder side, receives the plurality of data sequences on the decoder side, and uses the Key frame to form a Wyner-Ziv frame. 2 shows a system having a function of creating auxiliary information for decoding.

  The parallel DVC system decodes a Wyner-Ziv frame from a key frame, an encoder 30 having a channel encoder, an encoder 31 having an information source encoder, a decoder 32 having an information source decoder, and a key frame. Motion compensation unit 33 for performing auxiliary information necessary for motion compensation, bit likelihood estimation unit 34 for estimating bit likelihood from auxiliary information, decoding unit 35 having a channel decoder, and inverse quantization using auxiliary information Are constituted by a reconstruction unit 36, binary conversion units 37 and 38, and an inverse binary conversion unit 39.

  Each component will be described below.

  First, it is divided into a Key frame and a Wyner-Ziv frame from the information source, and the Wyner-Ziv frame is input to the binary conversion unit 27, and Gray Code (document “Gray, F.“ Pulse Code Communication. ”United States Patent Number. 2632058. March 17, 1953 ") is performed and sent to the encoding unit 30.

  In the encoding unit 30, an error correction code is used for each bit plane, and only the generated syndrome bits are transmitted simultaneously (in parallel) to the decoding unit 35 (decompression apparatus). On the other hand, the Key frame is compressed by the encoding unit 31 according to an international standard such as JPEG2000, and sent to the decoding unit 32 (decompression apparatus).

  The Key frame is decompressed by an international standard such as JPEG2000 in the decoding unit 32 (decompression apparatus) and divided into two paths. In one pass, it is output as a decoded image of a Key frame. The other path is sent to the motion compensation unit 23 (auxiliary information creation device). In the motion compensation unit 33, auxiliary information is created based on, for example, the above-described formula (1) so that the information is as equal as possible to the Wyner-Ziv frame.

  The created auxiliary information is sent to the binary conversion unit 38 and the reconstruction unit 36. The binary conversion unit 38 performs the same conversion as the binary conversion unit 37 and sends the same to the bit likelihood estimation unit 34.

  The bit likelihood estimator 34 estimates the bit likelihood of each bit plane by the following equation using the above equation (2).

上記の式（６）、式（７）は、前述の式（４）及び式（５）とは異なり、劣化による符号化効率の低下を防ぐために２値変換部３８においてGray Codeと呼ばれる２値変換を行うことにより、上位・下位ビット間の尤度推定誤差の影響を軽減している。

The above formulas (6) and (7) are different from the above formulas (4) and (5), and the binary conversion unit 38 uses a binary code called Gray Code in order to prevent a decrease in coding efficiency due to deterioration. By performing the conversion, the influence of the likelihood estimation error between the upper and lower bits is reduced.

  Using the bit likelihood derived from the equation (6) in the bit likelihood estimator 34, the decoding unit 35 performs a decoding process using the message probability propagation algorithm using the syndrome bits created by the encoder 30. Done to get a quantized Wyner-Ziv frame. The quantized Wyner-Ziv frame is transferred to the inverse binary conversion unit 39, and the binary conversion of the binary conversion units 37 and 38 is solved.

  Finally, the reconstruction unit 36 uses the auxiliary information generated by the motion compensation unit 33 and the Wyner-Ziv frame quantized by the decoding unit 35 output from the inverse binary conversion unit 29 as much as possible to the original Wyner. Reconstruction is performed so that it is close to a -Ziv frame, and a decoded Wyner-Ziv frame is output. This process is performed according to the above-described equation (3), for example.

  Through the above operation, parallel decoding processing of Wyner-Ziv frames can be performed, and high-speed decoding processing of a heavy message probability propagation algorithm can be expected.

[Third embodiment]
FIG. 6 shows the configuration of a parallel DVC system according to the third embodiment of the present invention.

  The parallel DVC system shown in the figure encodes a plurality of data strings (Key frame and Wyner-Ziv frame) on the encoder side, receives a plurality of data strings on the decoder side, and uses the Key frame to generate a Wyner-Ziv frame. 2 shows a system having a function of creating auxiliary information for decoding.

  The parallel DVC system is necessary for decoding a Wyner-Ziv frame from a Key frame, an encoder 40 having a communication encoder, an encoder 41 having an information source encoder, a decoder 42 having an information source decoder. Motion compensation unit 43 for performing auxiliary information by motion compensation, bit likelihood estimation unit 44 for estimating bit likelihood from auxiliary information, decoding unit 45 having a channel decoder, and inverse quantization using auxiliary information A reconfiguration unit 46 and a rate control unit 47 are included.

  Each component will be described below.

  The information source is divided into a Key frame and a Wyner-Ziv frame. The Key frame is input to the encoding unit 41, and the Wyner-Ziv frame is input to the encoding unit 40 and compressed. In the encoding unit 41 to which the key frame is input, compression is performed according to an international standard such as JPEG2000, and the compressed data is sent to the decoding unit 42 (decompression apparatus). On the other hand, in the encoding unit 40 to which the Wyner-Ziv frame is input, an error correction code is used for each bit plane. At that time, according to the bit amount estimated by the rate control unit 47, syndrome bits are sent in parallel to the decoding unit 45 (decompression apparatus). The rate control unit 47 estimates the bit amount necessary for decoding by the following equation.

但し、Ｐ^L _eは下記で与えられる。

However, P ^L _e is given by the following.

復号部３２（伸長装置）において、Ｋｅｙフレームは、ＪＰＥＧ２０００等の国際標準規格で伸長が行われ、２つのパスに分けられる。１つのパスではＫｅｙフレームの復号画像として出力され、もう一方のパスは補助情報作成部３３に送られる。補助情報作成部４３では、Wyner-Zivフレームとできるだけ等しい情報となるように、例えば、前述の式（１）に基づいて補助情報が作成される。

In the decoding unit 32 (decompression apparatus), the Key frame is decompressed according to an international standard such as JPEG2000 and divided into two paths. One pass is output as a decoded image of a Key frame, and the other pass is sent to the auxiliary information creation unit 33. In the auxiliary information creation unit 43, the auxiliary information is created based on, for example, the above-described formula (1) so that the information is as equal as possible to the Wyner-Ziv frame.

  The created auxiliary information is sent to the bit likelihood estimation unit 44 and the reconstruction unit 46. The bit likelihood estimator 44 estimates the bit likelihood of each bit plane by using the following equation (2).

但し、Ｚ_Ｌはあるビットが１を取る範囲を表しており、次式で与えられる。

However, Z _L represents a range where a certain bit takes 1 and is given by the following equation.

復号部４５では、上記の式（１０）により、導出されたビット尤度を用いて、符号化部４０で生成されたシンドロームビットを、メッセージ確率伝搬アルゴリズムにより復号し、量子化されたWyner-Zivフレームを得る。

In the decoding unit 45, the syndrome bit generated by the encoding unit 40 is decoded by the message probability propagation algorithm using the derived bit likelihood according to the above equation (10), and quantized Wyner-Ziv Get the frame.

  Finally, the reconstruction unit 46 uses the auxiliary information output from the motion compensation unit 43 and the quantized Wyner-Ziv frame output from the decoding unit 45 so as to be as close to the original Wyner-Ziv frame as possible. Reconfiguration is performed and a decoded Wyner-Ziv frame is output. This process is performed according to, for example, the equation (3).

  With the above operation, rate control on the encoder side can be performed in the parallel decoding process of the Wyner-Ziv frame.

<Other embodiments>
As another embodiment of the third embodiment, a method for further increasing the coding efficiency will be described.

  In the conventional DVC system, since it was not possible to estimate how much syndrome bits are required on the decoding side, sufficient syndrome bits were generated using a very large matrix, and decoding was performed using a part of them. The encoding efficiency is reduced. Therefore, a matrix of a size that will actually be used on the decoding side is used by using the syndrome bit rate estimation formula of the formula (8) described above.

  The encoding unit 40 calculates the bit amount obtained by the above equation (8).

を用いて、シンドロームビットを生成するための生成行列を決定し、また、復号部４５において、当該ビット量を用いて復号のための検査行列を決定する。符号化部４０における生成行列は、

Is used to determine a generation matrix for generating syndrome bits, and the decoding unit 45 determines a check matrix for decoding using the bit amount. The generator matrix in the encoding unit 40 is

で求められる行列である。符号化部４０は、生成された当該生成行列を用いてシンドロームビットを生成する。符号化部４０は、シンドロームビットを復号部４５に送信する。

This is the matrix obtained by. The encoding unit 40 generates syndrome bits using the generated generation matrix. The encoding unit 40 transmits the syndrome bits to the decoding unit 45.

  When receiving the syndrome bits from the encoding unit 40, the decoding unit 45 counts the number of the syndrome bits and generates a parity check matrix as follows.

syndrome bit, k + syndrome bit
(However, (syndrome bit) row, (k + syndrome bit) column)
The decoding unit 45 decodes the syndrome bits using the check matrix.

  However, the size of the generation matrix of the encoding unit 40 and the size of the check matrix of the decoding unit 45 are not fixed to the above-described size, and the encoding efficiency can be improved by using a matrix having a size close to the matrix. Is possible.

  Thus, by designing the LDPC using the bit amount of the rate control unit 47 in the encoding unit 40 and the decoding unit 45, it is possible to improve the encoding efficiency.

  In addition, for each of the configuration on the encoding side and the configuration on the decoder side, the operation of those components is constructed as a program and installed in a computer used as an encoding side device and a decoding side device for execution, or a network It is possible to circulate through.

  Further, the constructed program can be stored in a portable storage medium such as a hard disk, a flexible disk, or a CD-ROM, and can be installed or distributed in a computer.

  The present invention is not limited to the above-described embodiments and examples, and various modifications and applications can be made within the scope of the claims.

  The present invention can be applied to an encoding / decoding technique of an image encoding communication system.

It is a principle block diagram of this invention. [BRIEF DESCRIPTION OF THE DRAWINGS] It is a block diagram of the Wyner-Ziv encoding system used as the basis of this invention. It is a block diagram of the Wyner-Ziv encoding system of the spatial domain of this invention. It is a block diagram of the parallelization DVC system in 1st Example of this invention. It is a block diagram of the parallelization DVC system in the 2nd Example of this invention. It is a block diagram of the parallelization DVC system in the 3rd Example of this invention. It is a block diagram of conventional DSC (Distributted Source Coding). This is an example in which a turbo code is used for DVC (Distributed Video Coding).

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Communication path encoding part 2 Communication path decoding part 3 Reconstruction part 4 Bit likelihood estimation part 5 Auxiliary information preparation part 6 Information source encoding part 7 Information source decoding part 11 Quantization means, scalar quantizer 12 Encoding means , Rate adaptive LDPC encoder 13 decoding means, rate adaptive LDPC decoder 14 reconstruction means, reconstruction section 15 motion compensation means, motion compensator 20 encoding section 21 encoding section 22 decoding section 23 motion compensation section 24 bit likelihood Degree estimation unit 25 Decoding unit 26 Reconstruction unit 30 Encoding unit 31 Encoding unit 32 Decoding unit 33 Motion compensation unit 34 Bit likelihood estimation unit 35 Decoding unit 36 Reconstruction unit 37 Binary conversion unit 38 Binary conversion unit 39 Inverse Binary conversion unit 40 Encoding unit 41 Encoding unit 42 Decoding unit 43 Motion compensation unit 44 Bit likelihood estimation unit 45 Decoding unit 46 Reconfiguration unit 47 Rate control unit

Claims (10)

An image encoding / decoding system comprising an encoder side for encoding a moving image frame from an information source using DVC (Distributed Video Coding), and a decoder side for decoding the encoded frame,
The encoder side
When a Wyner-Ziv frame that is compressed without using the information of the preceding and following frames and is expanded using the preceding and following frames is input from the information source, quantization means that divides into bit planes;
Encoding means for generating syndrome bits for the bit plane and transmitting in parallel to the decoder side together with the bit plane;
Have
The decoder side
When a key frame that is compressed and decompressed without using the information of the previous and subsequent frames is input from the information source, auxiliary information in units of blocks is generated using motion compensation to decode the Wyner-Ziv frame. , Motion compensation means for obtaining a bit likelihood suitable for parallelization from the auxiliary information;
Decoding means for decoding the bit plane using the syndrome bits received from the encoding means on the encoder side and the bit likelihood;
Reconstructing means for reconstructing and outputting the bit plane decoded by the decoding means using the auxiliary information;
An image encoding / decoding system comprising: An image encoding / decoding system comprising an encoder side for encoding a moving image frame from an information source using DVC (Distributed Video Coding), and a decoder side for decoding the encoded frame,
The encoder side
A first encoding means for compressing and transmitting the key frame to the decoder side when a key frame to be compressed and decompressed without using information of previous and subsequent frames is input from the information source;
When a Wyner-Ziv frame that is compressed without using the information of the preceding and following frames and is expanded using the preceding and following frames is input from the information source, it is divided into bit planes, and an error correction code is assigned to each bit plane. Generating syndrome bits and transmitting only the syndrome bits in parallel to the decoder side, and
The decoder side
First decoding means for decompressing the compressed Key frame received from the first encoding means, dividing it into two paths, and outputting one as a decoded image of the Key frame;
When a path of another divided Key frame input from the first decoding means is input , a motion vector is generated from the Wyner-Ziv frame in units of blocks, and processing to match the range of the image is performed. Motion compensation means for creating auxiliary information for decoding the Wyner-Ziv frame ,
When the auxiliary information is input from the motion compensation means, bit likelihood estimation means for obtaining the bit likelihood of each bit plane;
Second decoding means for decoding a syndrome bit received from the second encoding means using a message probability propagation algorithm using the bit likelihood to obtain a quantized Wyner-Ziv frame;
Reconstructing means for reconstructing and outputting the quantized Wyner-Ziv frame by performing inverse quantization using the compensation information generated by the motion compensation means;
An image encoding / decoding system comprising: An image encoding / decoding system comprising an encoder side for encoding a moving image frame from an information source using DVC (Distributed Video Coding), and a decoder side for decoding the encoded frame,
The encoder side
A first encoding means for compressing and transmitting the key frame to the decoder side when a key frame to be compressed and decompressed without using information of previous and subsequent frames is input from the information source;
When a Wyner-Ziv frame that is compressed without using the information of the preceding and following frames and is decompressed using the preceding and following frames is input from the information source, binary conversion is performed, and the bit plane is divided. Second encoding means for generating syndrome bits using an error correction code every time and transmitting the syndrome bits to the decoder side in parallel;
First decoding means for decompressing the compressed Key frame received from the first encoding means, dividing it into two paths, and outputting one as a decoded image of the Key frame;
When the other path of the first decoding means is input , a motion vector is generated in units of blocks from the Wyner-Ziv frame , and the Wyner-Ziv frame is decoded by performing processing that matches the range of the image. and motion compensation means for generating an auxiliary information for creating auxiliary information for,
A bit likelihood estimating means for binary-converting the auxiliary information to obtain a bit likelihood of each bit plane;
Second decoding to obtain a quantized Wyner-Ziv frame by decoding the syndrome bits of the Wyner-Ziv frame transmitted from the second encoding means using a message likelihood propagation algorithm using the bit likelihood Means,
Reconstructing means for performing inverse binary transformation on the quantized Wyner-Ziv frame, performing inverse quantization using the auxiliary information generated by the motion compensation means, and outputting the result;
An image encoding / decoding system comprising: An image encoding / decoding system comprising an encoder side for encoding a moving image frame from an information source using DVC (Distributed Video Coding), and a decoder side for decoding the encoded frame,
The encoder side
A first encoding means for compressing and transmitting the key frame to the decoder side when a key frame to be compressed and decompressed without using information of previous and subsequent frames is input from the information source;
When a Wyner-Ziv frame that is compressed without using the information of the preceding and following frames and is decompressed using the preceding and following frames is input from the information source, rate control means for obtaining a bit amount necessary for decoding;
Second encoding means for dividing the Wyner-Ziv frame into bit planes, creating syndrome bits, and transmitting in parallel to the decoder side according to the bit amount required for the decoding obtained by the rate control means; ,
First decoding means for decompressing the compressed Key frame received from the first encoding means, dividing it into two paths, and outputting one as a decoded image of the Key frame;
When the other path of the first decoding means is input , a motion vector is generated in units of blocks from the Wyner-Ziv frame , and the Wyner-Ziv frame is decoded by performing processing that matches the range of the image. and motion compensation means for generating an auxiliary information for,
Bit likelihood estimation means for obtaining a bit likelihood of each bit plane using the auxiliary information;
Second decoding to obtain a quantized Wyner-Ziv frame by decoding the syndrome bits of the Wyner-Ziv frame transmitted from the second encoding means using a message likelihood propagation algorithm using the bit likelihood Means,
Reconstructing means for dequantizing and outputting the quantized Wyner-Ziv frame using the auxiliary information generated by the motion compensation means;
An image encoding / decoding system comprising: The second encoding means includes
Means for generating a generation matrix for generating syndrome bits using the bit amount obtained in the rate control step, generating syndrome bits using the generation matrix, and transmitting the generation bits to the second decoding means; Including
The second decoding means includes
5. The means for generating a check matrix based on the number of bits of the syndrome bits transmitted from the second encoding means or the estimated number of bits and decoding the syndrome bits using the check matrix. Image encoding / decoding system. Image encoding / decoding in an image encoding / decoding system comprising an encoder side that encodes a moving image frame from an information source using DVC (Distributed Video Coding) and a decoder side that decodes the encoded frame A method,
On the encoder side,
When the Wyner-Ziv frame compressed by the quantization means without using the information of the preceding and following frames and expanded using the preceding and following frames is input from the information source, it is divided into bit planes,
The encoding means generates a syndrome bit for the bit plane, and transmits the syndrome bit together with the bit plane to the decoder side,
On the decoder side,
When a key frame that is compressed and decompressed without using information of previous and subsequent frames by the motion compensation means is input from the information source, in order to decode the Wyner-Ziv frame, block compensation is performed using motion compensation. Auxiliary information is generated, a bit likelihood suitable for parallelization is obtained from the auxiliary information,
The decoding means decodes the bit plane using the syndrome bits received from the encoding means on the encoder side and the bit likelihood,
An image encoding / decoding method, wherein a decoding bit plane is reconstructed by the reconstruction means using the auxiliary information and output. Image encoding / decoding in an image encoding / decoding system comprising an encoder side that encodes a moving image frame from an information source using DVC (Distributed Video Coding) and a decoder side that decodes the encoded frame A method,
On the encoder side,
When a key frame that is compressed and decompressed without using the information of the preceding and following frames by the first encoding means is input from the information source, the key frame is compressed and transmitted to the decoder side,
When the Wyner-Ziv frame compressed by the second encoding means without using the information of the preceding and following frames and decompressed using the preceding and following frames is input from the information source, it is divided into bit planes, Generating syndrome bits using an error correction code for each bit plane, and transmitting only the syndrome bits in parallel to the decoder side,
On the decoder side,
The first decoding means decompresses the compressed Key frame received from the first encoding means, divides it into two paths, and outputs one as a decoded image of the Key frame,
When the motion compensator receives the path of the other key frame divided from the first decoder, the motion compensator creates a motion vector in units of blocks from the Wyner-Ziv frame, and adds the motion vector to the image range. Generate auxiliary information for decoding the Wyner-Ziv frame by performing the matching process ,
When the auxiliary information is input from the motion compensation unit by the bit likelihood estimation unit, the bit likelihood of each bit plane is obtained,
The second decoding means decodes the syndrome bits received from the second encoding means by the message probability propagation algorithm using the bit likelihood, and obtains a quantized Wyner-Ziv frame;
Reconstructing means reconstructs the quantized Wyner-Ziv frame by performing inverse quantization using the compensation information generated by the motion compensation means, and outputs the image code. / Decoding method. Image encoding / decoding in an image encoding / decoding system comprising an encoder side that encodes a moving image frame from an information source using DVC (Distributed Video Coding) and a decoder side that decodes the encoded frame A method,
On the encoder side,
When a key frame that is compressed and decompressed without using the information of the preceding and following frames by the first encoding means is input from the information source, the key frame is compressed and transmitted to the decoder side,
When the Wyner-Ziv frame compressed by the second encoding means without using the information of the preceding and following frames and expanded using the preceding and following frames is input from the information source, binary conversion is performed, Dividing into planes, creating syndrome bits using an error correction code for each bit plane and transmitting them in parallel to the decoder side,
The first decoding unit decompresses the compressed key frame received from the first encoding unit, divides the compressed key frame into two paths, outputs one as a decoded image of the key frame, and the other as motion compensation. Output to the means,
When the other path is input from the first decoding unit, the motion compensation unit creates a motion vector in units of blocks from the Wyner-Ziv frame , and performs processing to match the range of the image. -Generate auxiliary information to create auxiliary information for decoding the Ziv frame ,
The auxiliary likelihood information is subjected to binary conversion by a bit likelihood estimating means to obtain a bit likelihood of each bit plane,
The Wyner-Ziv, which is decoded by the message probability propagation algorithm using the bit likelihood with respect to the syndrome bits of the Wyner-Ziv frame transmitted from the second encoding means by the second decoding means, and is quantized. Get the frame,
An image which is subjected to inverse binary transformation on the quantized Wyner-Ziv frame by a reconstructing means, and is subjected to inverse quantization using the auxiliary information generated by the motion compensation means, and output. Encoding / decoding method. Image encoding / decoding in an image encoding / decoding system comprising an encoder side that encodes a moving image frame from an information source using DVC (Distributed Video Coding) and a decoder side that decodes the encoded frame A method,
On the encoder side,
When a key frame that is compressed and decompressed without using information of preceding and following frames by the first encoding means is input from the information source, the key frame is compressed and transmitted to the decoder side. An encoding step;
Rate control that obtains the bit amount necessary for decoding when a Wyner-Ziv frame that is compressed without using the information of the previous and subsequent frames and is expanded using the previous and subsequent frames is input from the information source. Steps,
The second encoding means divides the Wyner-Ziv frame into bit planes, creates syndrome bits, and transmits them in parallel to the decoder side according to the bit amount required for the decoding obtained in the rate control step. Performing a second encoding step,
On the decoding side,
The first decoding means decompresses the compressed Key frame received from the first encoding means, divides it into two paths, outputs one as a decoded image of the Key frame, and sets the other path as A first decoding step for outputting to the motion compensation means;
When the other path is input from the first decoding unit by the motion compensation unit, a motion vector is generated from the Wyner-Ziv frame in units of blocks, and processing is performed according to the range of the image. A motion compensation step for generating auxiliary information for generating auxiliary information for decoding a -Ziv frame ;
A bit likelihood estimating step of obtaining a bit likelihood of each bit plane using the auxiliary information by a bit likelihood estimating means;
The Wyner-Ziv, which is decoded by the message probability propagation algorithm using the bit likelihood with respect to the syndrome bits of the Wyner-Ziv frame transmitted from the second encoding means by the second decoding means, and is quantized. A second decoding step to obtain a frame;
A reconstruction step of performing inverse quantization on the quantized Wyner-Ziv frame using the auxiliary information generated by the motion compensation unit and outputting the quantized Wyner-Ziv frame;
An image encoding / decoding method characterized by: In the second encoding step, a generation matrix for generating syndrome bits is generated using the bit amount obtained in the rate control step, and syndrome bits are generated using the generation matrix, To the second decryption means,
In the second decoding step, a check matrix is generated based on the number of bits or the estimated number of the syndrome bits transmitted from the second encoding means , and the syndrome bits are decoded using the check matrix The image encoding / decoding method according to claim 9.

Images