KR20090045558A - Apparatus and method of decompressing distributed video coded video using error correction - Google Patents

Abstract

The present invention relates to an apparatus and method for decoding a distributed video encoded image using error correction. A decoding apparatus according to the present invention includes a key picture decoding unit for reconstructing a key picture transmitted from an encoding device; An auxiliary information generator configured to generate auxiliary information using the key picture reconstructed by the key picture decoder; A channel code decoder for estimating a quantized value using the parity bits transmitted from the encoding apparatus and the auxiliary information; An image restoring unit for restoring a WZ picture to be decoded by using the quantized value estimated by the channel code decoding unit and the auxiliary information; Determining whether a channel code decoding error occurs on the WZ picture by using the auxiliary information and the key picture reconstructed by the key picture decoding unit, and correcting an error of the reconstructed WZ picture based on a picture similarity. It is characterized by including the error correction. Accordingly, the subjective image quality of the reconstructed image can be significantly improved by correcting the decoding error occurring in the reconstructed image.

Description

Device for decoding distributed video encoded video using error correction and its method {APPARATUS AND METHOD OF DECOMPRESSING DISTRIBUTED VIDEO CODED VIDEO USING ERROR CORRECTION}

The present invention relates to an apparatus and a method for decoding a distributed video encoded image using error correction. The present invention relates to an error correction method for determining a decoding error of a reconstructed image and correcting the determined error to improve subjective image quality of the reconstructed image. The present invention relates to a distributed video coded video decoding apparatus and a method thereof.

Digital video data used in video conferencing, video on demand (VOD) receivers, digital broadcast receivers and cable television (CATV) is generally compressed by an efficient compression method rather than being used as it has a significant amount of data.

Compression standards such as MPEG, H.26x, etc. are used for many applications such as video player, VOD, video telephony, DMB, etc. Recently, due to the development of wireless communication such as 2.5G / 3G, It is also used for image transmission in a wireless mobile base.

Digital image data compression mainly uses three methods: reducing temporal redundancy, reducing spatial redundancy, and reducing statistical redundancy of generated codes. Among them, a representative method of reducing temporal redundancy is motion prediction and compensation technology.

Current coding techniques have achieved high coding efficiency by eliminating this temporal redundancy, but since the motion calculation and compensation technique which occupies the most computational amount in the video encoder also reduces the complexity of the encoder in a limited resource environment such as a sensor network. This is an important technical issue.

Distributed source coding (DSC) technology based on the Slepian-Wolf theory has attracted attention as one of methods for solving the complexity problem of the encoder. The Slepian-Wolf theory mathematically proves that even if the correlated sources are encoded independently, if the decoding is linked to each other, the coding gain can be as high as that obtained by predictive coding of the respective sources together.

The Distributed Video Coding (DVC) technology is a method of lossless compression that uses the lossless compression of the distributed source coding technology. The lossy coding of the Slepian-Wolf theory, which is also a theoretical foundation of the distributed source coding technology, is used. It is based on the Wyner-Ziv theory, which was expanded into. From the point of view of video encoding technology, both of these techniques mean that processing procedures such as motion prediction and compensation, which have been performed to reduce the redundancy between pictures, can be moved to the decoder without any loss of encoding gain.

Known techniques for distributed video coding include A. Aaron, S. Rane, R. Zhang, and B. Girod et al. In Proc. Wyner-Ziv coding technology based on a paper presented at the IEEE Data Compression Conference, 2003, "Wyner-Ziv coding for video: Applications to compression and error resilience." The distributed video encoding technique uses the similarity between neighboring pictures in a decoder to generate supplemental information about a current picture. The supplemental information is regarded as a virtual channel noise is added to the current picture to be reconstructed. The current picture is reproduced by removing noise in the auxiliary information using the channel code.

1 is a diagram illustrating a configuration of an encoder 110 and a decoder 130 corresponding to the conventional Wyner-Ziv coding technique.

As shown in FIG. 1, the encoder 110 according to the conventional Wyner-Ziv coding technique includes a key picture encoder 114, a block unitization unit 111, a quantization unit 112, and a channel code encoding unit 113. The decoder 130 corresponding thereto includes a key picture decoder 133, a channel code decoder 131, an auxiliary information generator 134, and an image restorer 132.

The encoder 110 according to the Wyner-Ziv coding technique classifies a picture to be encoded into two types. One is a picture to be encoded by the distributed video coding method (hereinafter referred to as a 'WZ picture'), and the other is a picture to be encoded by a conventional coding method instead of the distributed video coding method (hereinafter, referred to as a 'key picture'). )to be.

The key pictures are usually encoded by the key picture encoder 114 and transmitted to the decoder 130 in a predetermined method selected by a user, such as H.264 / AVC. The key picture decoder 133 of the decoder 130 corresponding to the encoder 110 according to the conventional Wyner-Ziv coding technique reconstructs key pictures encoded and transmitted by a predetermined method, and the auxiliary information generator 134. ) Generates side information corresponding to the WZ picture using the key picture reconstructed by the key picture decoding unit 133.

Typically, the auxiliary information generator 134 generates side information corresponding to the WZ picture to be restored by using interpolation method assuming linear motion between key pictures located before and after the WZ picture. In some cases, interpolation may be used. However, interpolation is used in most cases because interpolation is superior to interpolation in performance.

Meanwhile, in order to encode the WZ picture, the block unit unit 111 of the encoder 110 divides the input WZ picture into predetermined coding units, and the quantization unit 112 performs quantization for each coding unit. The channel code encoder generates a parity bit for the quantized value using the channel code.

The generated parity bits are stored in a parity buffer (not shown) and sequentially transmitted according to a request of the decoder 130 through a feedback channel. The channel code decoder 131 of FIG. 1 receives the parity transmitted from the encoder 110 and estimates the quantized value. The image restorer 132 of FIG. 1 receives the quantized value estimated by the channel code decoder 131 and inversely quantizes the reproduced WZ picture.

Ambiguity generated during inverse quantization in the above process is solved by referring to auxiliary information input by the auxiliary information generator 134. For a detailed description, see A. Aaron, S. Rane, R. Zhang, and B. Girod et al. In Proc. See "Wyner-Ziv coding for video: Applications to compression and error resilience" published in the IEEE Data Compression Conference, 2003.

FIG. 2 is a diagram illustrating a configuration of a channel code decoder 131 in the Wyner-Ziv coding technique of FIG. 1. As shown in FIG. 2, the channel code decoder 131 includes two SISO decoders 210a and 210b, an interleaver 213a and 213b, an inverse interleaver 214a and 214b, and a channel probability value calculator 211a and 211b), a determination unit 216 and a demux 215 (DEMUX).

The parity bit, which is composed of parity for the quantized value and parity for the interleaved value, and is transmitted from the encoder 110, is separated by the demux 215 (DEMUX) of FIG. 211b). Each channel probability calculator 211a or 211b calculates a channel probability value by receiving auxiliary information, probability statistical characteristics for noise, and parity bits transmitted from the encoder 110.

Each of the SISO decoders 210a and 210b performs decoding based on this channel probability value and an APrP (A Priori Probability) value provided from another SISO decoder 210a or 210b. Here, each of the SISO decoders 210a and 210b proceeds from the first state to the last state on the grid and obtains the forward state metric from the transition metric, and when the last state is reached, the SISO decoder 210a and 210b obtains the reverse state metric. The A Posteriori Probability (APoP) value and the extrinsic probability value are calculated using the state metric value and the transition metric value thus obtained.

The determination unit 216 calculates an error rate from the APoP, and if the value falls below the threshold value, the decoding ends. Otherwise, the other SISO decoders 210a and 210b repeat the same process. However, even after a predetermined repetition, if the error rate does not fall below the threshold, the decoder 130 may make an additional parity request to the encoder 110 through a feedback channel.

This decoding method basically corrects the noise in the auxiliary information using a channel code. However, because the motion is complicated and it is difficult to make the auxiliary information accurately, the noise in the auxiliary information is greatly increased. As a result, if the amount of parity bits transmitted is insufficient, an error of estimating an incorrect transition process is corrected.

Meanwhile, since the two SISO decoders 210a and 210b exchange extrinsic probabilities with each other, an incorrectly estimated transition result is transferred to another SISO decoder 210a and 210b to cause another error. Using the conventional method, this problem is inevitable unless a sufficient parity bit is received, and the value restored due to the cumulative error is sometimes completely different from the original value. Since noise is generated, there is a problem of significantly lowering the subjective image quality of an image.

Therefore, there is an urgent need for a technique for preventing channel code decoding errors below a predetermined value or correcting the errors by using peripheral information.

In order to meet the above technical requirements, the present invention determines whether a decoding error of a channel code has occurred in a reconstructed image using temporal and spatial similarity, and selectively corrects the error according to the result. An object of the present invention is to provide a method and apparatus for decoding a distributed video encoded image using error correction to improve the image quality of the image.

According to an aspect of the present invention, there is provided a decoding apparatus for a distributed video encoded image using error correction, the apparatus comprising: a key picture decoder for reconstructing a key picture transmitted from the encoding apparatus; An auxiliary information generator configured to generate auxiliary information using the key picture reconstructed by the key picture decoder; A channel code decoder for estimating a quantized value using the parity bits transmitted from the encoding apparatus and the auxiliary information; An image restoring unit for restoring a WZ picture to be decoded by using the quantized value estimated by the channel code decoding unit and the auxiliary information; Determining whether a channel code decoding error has occurred on the WZ picture by using the auxiliary information and the key picture reconstructed by the key picture decoding unit, and correcting an error of the reconstructed WZ picture based on a picture similarity. It is achieved by a decoding apparatus characterized by including error correction.

Here, the error correction unit may include: a decoding error determination unit for determining a correction target pixel in which a channel code decoding error occurs among pixels in the restored WZ picture based on the auxiliary information and the restored key picture; And a decoding error correcting unit configured to correct the correcting pixel based on a similarity between the correcting pixel and the neighboring pixel temporally and / or spatially determined by the decoding error determining unit.

The decoding error determination unit may include a spatial similarity measuring unit measuring a spatial similarity between a specific pixel in the reconstructed WZ picture and a neighboring pixel, and a temporal similarity measuring unit measuring a temporal similarity between the auxiliary information and the reconstructed WZ picture. At least one of; Comparing at least one of the spatial similarity and the temporal similarity with a preset threshold value, and determining that the correction target pixel exists when at least one of the spatial similarity and the temporal similarity is larger or smaller than the threshold value. It may include a final discrimination unit.

The spatial similarity measurer may measure the spatial similarity using a difference in pixel values between the specific pixel and the neighboring pixel.

The temporal similarity measurer may measure the temporal similarity using a difference in pixel values between the pixels corresponding to the restored WZ picture and the auxiliary information.

The decoding error correction unit may further include a spatial candidate value estimator estimating a spatial candidate value based on a spatial similarity between the correction target pixel and a neighboring pixel, and a temporal relationship between the correction target pixel and a corresponding pixel in the reconstructed key picture. At least one of a temporal candidate value estimator for estimating a temporal candidate value based on the similarity degree; And a final corrector configured to correct the pixel to be corrected using at least one of the temporal candidate value and the spatial candidate value.

Here, the spatial candidate value estimator may estimate a median value of pixel values of neighboring pixels of the correction target pixel as the spatial candidate value.

The temporal candidate value estimator may estimate the temporal candidate value through motion prediction for the reconstructed WZ picture by using at least one key picture reconstructed by the key picture decoder as a reference image.

Meanwhile, according to another embodiment of the present invention, there is provided a decoding method of a distributed video encoded image using error correction, the method comprising: (a) restoring a key picture transmitted from an encoding device; (b) generating auxiliary information using the reconstructed key picture; (c) estimating a quantized value using the auxiliary information and a parity bit transmitted from the encoding apparatus; (d) restoring a WZ picture to be decoded using the estimated quantized value and the auxiliary information; (e) determining whether a channel code decoding error has occurred on the WZ picture using the auxiliary information and the key picture reconstructed by the key picture decoding unit; and (f) correcting the error of the reconstructed WZ picture based on the picture similarity.

The step (e) may include measuring at least one of spatial similarity between a specific pixel in the reconstructed WZ picture and a neighboring pixel, and measuring temporal similarity between the auxiliary information and the reconstructed WZ picture. Wow; Comparing at least one of the spatial similarity and the temporal similarity with a preset threshold; And determining that there is a correction target pixel in which a channel code decoding error occurs in the reconstructed WZ picture, wherein at least one of the spatial similarity and the temporal similarity is greater than or less than the threshold value.

The spatial similarity may be measured based on a difference in pixel values between the specific pixel and the neighboring pixel.

Here, the temporal similarity may be measured based on a difference in pixel values between the pixels corresponding to each other between the reconstructed WZ picture and the auxiliary information.

And (f) estimating a spatial candidate value based on spatial similarity between the correction target pixel and a neighboring pixel, and based on temporal similarity between the correction target pixel and a corresponding pixel in the reconstructed key picture. Estimating a temporal candidate value; And correcting the correction target pixel using at least one of the temporal candidate value and the spatial candidate value.

Here, the spatial candidate value may be estimated as an intermediate value of pixel values of neighboring pixels of the correction target pixel in the reconstructed WZ picture.

The temporal candidate value may be estimated through motion prediction of the reconstructed WZ picture by using at least one key picture reconstructed by the key picture decoder as a reference image.

According to the present invention, it is possible to determine whether a decoding error occurs in a pixel based on a temporal similarity measurement with a neighboring pixel, a spatial similarity measurement, and a predetermined discrimination criterion in each pixel of the reconstructed image. In addition, only a pixel determined to have an error may be estimated by using a temporal and / or spatial similarity with a neighboring pixel to correct a candidate value. Therefore, the subjective image quality of the reconstructed image can be significantly improved by correcting the decoding error occurring in the reconstructed image according to the present invention.

Hereinafter, with reference to the accompanying drawings will be described in detail the present invention.

3 is a diagram illustrating a configuration of a Wyner-Ziv encoding apparatus 10 and a decoding apparatus 30 of a distributed video encoded image using error correction including decoding error determination and correction according to the present invention.

Referring to FIG. 3, the Wyner-Ziv encoding apparatus 10 according to the present invention includes a key picture encoder 12 and a WZ picture encoder 11. In addition, the decoding apparatus 30 according to the present invention includes a key picture decoding unit 33, a channel code decoding unit 32, an auxiliary information generating unit 35, an image restoring unit 34, and an error correction unit 36. Include.

The key picture decoder 33 restores the key picture using the information received from the key picture encoder 12, and the auxiliary information generator 35 assists with the current WZ picture to be reproduced using the restored key pictures. Generate information.

The channel code decoder 32 estimates the quantized values by using the auxiliary information input from the auxiliary information generator 35 and the parity bits received from the Wyner-Ziv encoding apparatus 10. The image restorer 34 reconstructs the WZ picture using the quantized value and the auxiliary information estimated by the channel code decoder 32.

Here, the WZ picture a restored by the image restoring unit 34 is input to the error correcting unit 36, and the error correcting unit 36 is provided with the auxiliary information (b) from the auxiliary information generating unit 35. The reconstructed WZ with the image quality significantly improved by determining and correcting a position where a channel code decoding error occurs on the reconstructed WZ picture a by using the key pictures c reconstructed by the key picture decoding unit 33. Reproduce the picture (d).

When the channel code decoding unit 32 of FIG. 3 performs channel code decoding and determines that a reliable quantized value cannot be estimated, the channel code decoding unit 32 continuously transmits a parity bit within a predetermined limit until a reliable estimation is possible. It is configured to request and receive the Ziv encoding apparatus 10.

In this case, only the amount of parity required for decoding is transmitted from the Wyner-Ziv encoding apparatus 10, which is efficient in terms of bit rate reduction. This is only possible if there is a reverse channel (ie a feedback channel) requesting the parity bits. In order to alleviate this problem, according to the user's configuration, the channel code decoding unit 32 receives a predetermined amount of parity bits in advance without requesting a parity request in advance and does not request parity on the reverse channel while using the same. Can be.

Also in the above case, after exhausting all information on the sent parity bits, if it is determined that the reliability is still low, it may be configured to further send information about the parity bits. In addition, assuming that the reverse channel is not used, the Wyner-Ziv encoding apparatus 10 always transmits a predetermined amount of parity to the decoding apparatus 30, and the decoding apparatus 30 is configured to not require a parity bit. It may be.

In addition, it is preferable that the channel code used in the channel code decoder 32 of FIG. 3 use a turbo code that is said to almost reach the Shannon limit or use an LDPC channel code. It is also apparent that other channel codes with good coding efficiency and error correction can be used.

4 is a diagram showing the configuration of the error correction unit 36 according to the present invention. As shown in FIG. 4, the error correction unit 36 according to the present invention includes a decoding error determination unit 361 and a decoding error correction unit 362.

The decoding error determining unit 361 determines whether a channel code decoding error has occurred at each position of the WZ picture a restored by the image restoring unit 34. An example of the configuration is shown in FIG. As it is. The decoding error correction unit 362 corrects a pixel in which an error occurs by using a similarity with spatial or temporal or both peripheral information. An example of the configuration is illustrated in FIG. 6.

Referring to FIG. 5, the decoding error determination unit 361 measures spatial similarity between the restored pixels in the WZ picture (a), which is the reconstructed image input by the image restoration unit 34, and the neighboring pixels. Temporal similarity measuring unit for measuring the temporal similarity between the measuring unit 361a, the auxiliary information (b) input from the auxiliary information generating unit 35, and the WZ picture (a) pixel, which is a reconstructed image from the image restoring unit 34, 361c, and whether there is a channel code decoding error at a corresponding position in the WZ picture (a) which is a reconstructed image from the measurement results of the spatial similarity measuring unit 361a and the temporal similarity measuring unit 361c. And a final judging section 361b for judging. Here, the temporal similarity measurer 361c may be configured to use the key picture c generated by the key picture decoder 33 as shown in FIG. 5.

The spatial similarity measuring unit 361a of the decoding error determining unit 361 determines the similarity between the reconstructed pixel in the reconstructed image WZ picture a and its neighboring pixels and the neighboring pixel shown in FIG. 7. The difference between the maximum value, the minimum value, and the WZ picture (a) which is a reconstructed image is calculated as shown in [Equation 1].

[Equation 1]

는 (i,j) 위치에서의 복원영상인 WZ픽춰(a)내의 해당 화소값이며,

은 도 7에 도시된 바와 같이, 이에 대한 복원영상인 WZ픽춰(a)내 공간적 주변 화소값들이다.here,

Is the corresponding pixel value in the WZ picture (a) which is the reconstructed image at position (i, j),

7 are spatial peripheral pixel values in the WZ picture (a) which is a reconstructed image thereof.

을 계산한다. 여기서,

는 (i,j) 위치에서의 보조정보(a) 내 화소값이다.Meanwhile, the temporal similarity measurer 361c of the decoding error determiner 361 corresponds to the difference between the auxiliary information (b) and the pixel of the WZ picture (a), which is a reconstructed image.

Calculate here,

Is the pixel value in the auxiliary information (a) at the position (i, j).

The above equations used for the spatial and temporal similarity measurement are one example for measuring the spatial or temporal similarity, and various approaches are possible in addition to the above embodiments.

을 오류판별기준 정보인 소정의 임계값 A와 비교하여, 그 값이 임계값 A 보다 크면 해당 화소에 복호화 오류가 발생했을 여지가 큰 것으로 판별한다. 이는 보조정보(b)와 현재픽춰의 유사도는 일반적으로 높으므로 복원영상인 WZ픽춰(a)의 화소값이 보조정보(b)와 어느 정도 이상으로 값 차이가 나면 해당 화소에 복호화 오류가 발생했을 여지가 클 것이라는 일반적 성질에 의거한다.The final judging unit 361b finally determines whether a channel code decoding error has occurred based on either or both of temporal and spatial similarities. For example, the final determining unit 361b calculates the calculated

Is compared with a predetermined threshold value A, which is the error discrimination reference information, and if the value is larger than the threshold value A, it is determined that a decoding error is likely to occur in the pixel. Since the similarity between the auxiliary information (b) and the current picture is generally high, if the pixel value of the WZ picture (a), which is a reconstructed image, differs from the auxiliary information (b) by more than a certain value, a decoding error may have occurred in the corresponding pixel. Based on the general nature that the room is large.

또는

값이 이에 해당하는 오류판별 기준정보인 소정의 임계값 B 보다 크면 해당화소의 위치에 복호화 오류가 발생한 것으로 판별한다. 복호화 오류 판별부의 구현에는 여러 방법이 있을 수 있으며, 이에 설명한 어느 방법에 의해 본 특허가 제한받는 것은 아니다. Afterwards, for a pixel that is likely to have a decoding error, Equation 1

or

If the value is larger than the predetermined threshold value B, which is corresponding error discrimination reference information, it is determined that a decoding error has occurred at the position of the corresponding pixel. There may be various methods for the implementation of the decoding error determination unit, and the present patent is not limited by any of the methods described herein.

Referring to FIG. 6, the decoding error correction unit 120 performs channel decoding error correction by using all or part of the information shown in FIG. 8 with respect to the position e determined that there is a decoding error.

When calculating a value for correcting the channel decoding error, it may be implemented in any one or a combination thereof shown in Cases 1 to 7 shown in FIG. 8, and the input data required by the decoding error correction unit is determined according to this form. do. The decoding error correcting unit corrects the temporal similarity between the spatial candidate value estimator 362a estimating a reasonable candidate value based on the spatial similarity with neighboring pixels, and the reconstructed key picture c and the reconstructed image WZ picture a. And a temporal candidate value estimator 362c estimating a reasonable candidate value based on the result, and a final corrector 362b correcting a pixel in which an error occurs using the estimated candidate values.

As illustrated in FIG. 8, the decoding error correction unit 362 may be configured in various forms according to what image information is used to calculate a correction value. That is, when using the case 1 of FIG. 8 considering only the spatial candidate value, the temporal candidate value estimator 362c constituting the decoding error correction unit may be omitted in the configuration.

는 보정된 복원영상인 WZ픽춰(a)의 화소값이다An example of the spatial candidate value estimator 362a of the decoding error corrector 362 is to estimate a median value with neighboring pixels shown in FIG. 7 as the most suitable candidate value using Equation 2. This corresponds to Case 1 of FIG. 8. Here, in Equation 2, the intermediate value is not the only method, and various functions can be used according to the user. In [Equation 2]

Is the pixel value of the WZ picture (a) that is the corrected reconstructed image.

[Equation 2]

The temporal candidate value estimator 362c illustrated in FIG. 6 may be configured to add and use an auxiliary image b generated by the auxiliary information generator 35 according to an implementation, as indicated by a dotted line in FIG. 6. .

The spatial candidate value estimator 362a and the temporal candidate value estimator 362c are paralleled so that the final correction unit 362b uses the estimation results of the spatial candidate value estimator 362a and the temporal candidate value estimator 362c. It may have a general configuration, and may be configured in a sequential structure for using one of the estimation results in the other. In addition, various configurations can be considered because an approach of using only one candidate value estimator can be used for the purpose of simplifying the apparatus.

The temporal candidate value estimator 362c of the decoding error correcting unit reconstructs the candidate most similar to the corresponding position of the WZ picture a, which is the reconstructed image, as shown in FIG. 8. , c2, c3) are found through motion prediction using the reference image. The reference image used for such motion prediction may be a previous key picture in time based on the WZ picture to be corrected and reconstructed (this is called forward prediction, and Case 3 of FIG. 8), or a key picture thereafter. It may be referred to as backward prediction, and Case 4 of FIG. 8 corresponds to this), and may be a case of using both pre- and post-key pictures (this is referred to as bidirectional prediction, and Case 5 or Case 7 of FIG. 8 corresponds to this).

Here, the reference image position most similar to the current position of the WZ picture (a), which is the reconstructed image, is estimated by searching on the selected key picture. Such forward or reverse motion prediction or bidirectional prediction is performed to find three candidate positions having the highest similarity in each direction, and estimate the candidate position having the smallest SAD among them.

Alternatively, it is possible to approach the direction of generating a reasonable candidate value by combining all or part of the three candidate values through a predetermined procedure without selecting one of the three found in the above process. In addition, a configuration of estimating candidate values by performing only one of forward and backward motion prediction for the purpose of reducing the amount of motion prediction may be possible. In addition, since the motion prediction of various configurations can be considered, the scope of the present invention includes all reasonable motion prediction configurations.

The final corrector 362b finally corrects the value of each position determined that an error of the WZ picture a, which is the reconstructed image, has occurred, to the calculated candidate value.

In order to further increase the accuracy of error correction through the temporal candidate value estimation described above, the decoding error correction unit 362 of FIG. 4 may be configured as shown in FIG. 9. Referring to FIG. 9, the decoding error corrector 362 ′ uses spatial information as the spatial candidate value estimator 362a of FIG. 6 performs through the spatial candidate value estimator and corrector 362 ′ a. After the correction candidate value is obtained and the error occurrence position value is corrected, the motion predictor 362'b performs the temporal candidate value prediction through the motion prediction described above, and the final corrector 362'c uses the value. Correct the error location value again. The spatial candidate value estimator and correction unit 362'a can further improve error correction and use this value to predict motion, so that more accurate candidate values can be estimated when temporal candidate values are predicted. Do.

Hereinafter, a channel decoding error detection and correction method of a distributed video encoded image using error correction according to the present invention will be described in detail with reference to FIGS. 10 to 13.

The channel decoding error detection step S10 may be implemented in various ways, but an embodiment thereof is as shown in FIG. Referring to FIG. 11, the similarity between each position of the WZ picture a which is a reconstructed image and neighboring pixels is calculated in the spatial similarity measuring step S11. This spatial similarity measurement calculates the difference between each pixel value of the WZ picture (a), which is a reconstructed image, and the maximum value and the minimum value among neighboring pixel values as shown in [Equation 1].

In the temporal similarity measuring step (S12), the difference between the auxiliary information (b) and the pixel of the WZ picture (a), which is a reconstructed image, is calculated as temporal similarity.

Calculating the difference between the values to measure the similarities is only one example, and other calculation schemes may be used.

,

값 중 하나가 소정의 임계값 B 보다 크면 해당 화소에 복호화 오류가 발생한 것으로 판별한다.The spatial and temporal similarity calculated by the step S11 and the step S12 is compared with the predetermined threshold value A by comparing the measured temporal similarity value with the predetermined threshold A, as described above, in the error determination step S13. Greater than A and at the same time the measured spatial similarity value, i.e.,

,

If one of the values is larger than the predetermined threshold B, it is determined that a decoding error has occurred in the corresponding pixel.

In the embodiment of the channel decoding error detection step (S10) of FIG. 10 shown in FIG. 11, the configuration using both spatial or temporal similarities in the error determination step (S13) is shown. As described above, step S13 may be performed to consider only one of spatial or temporal similarities.

In the channel decoding error correction step S30 shown in FIG. 10, the channel decoding is performed using all or part of the information shown in FIG. 8 with respect to the position where the channel decoding error is determined by the channel decoding error detection step S10. Perform error correction.

When calculating the value to correct the channel decoding error may be implemented in any one or a combination thereof as shown in Case 1 to Case 7 shown in Figure 8, according to this form required in the decoding error correction step (S30) The data is determined. As shown in FIG. 12, for estimating the channel decoding error, a spatial candidate value estimating step (S31) of estimating a reasonable candidate value based on spatial similarity with neighboring pixels, a reconstructed key picture (c) and a reconstructed image ( A temporal candidate value estimating step (S32) of estimating a reasonable candidate value based on the temporal similarity between a) and a final correction step (S33) of correcting a pixel in which an error occurs using the estimated candidate values are performed. In the channel decoding error correction step (S30), as shown in FIG. 8, various types of data may be used depending on which image information is used to calculate a correction value. In the case of the case 1 of FIG. 8 considering only the spatial candidate value, the temporal candidate value estimating step S32 may be omitted in the channel decoding error correction step S30 of FIG. 12.

Meanwhile, the spatial candidate value estimating step S31 and the temporal candidate value estimating step S32 may be performed in a parallel structure or a sequential structure according to a user's configuration, or only one of them may be performed to obtain convenience of the step.

An embodiment of performing the spatial candidate value estimating step (S31) is to estimate a median value with neighboring pixels shown in FIG. 7 as the most suitable candidate value by using Equation (2). This corresponds to Case 1 of FIG. 8. In this equation, the intermediate value is not the only method in Equation 2, and various functions can be used according to the user.

The spatial candidate value estimating step S31 and the temporal candidate value estimating step S32 are performed in parallel before the final correction step S33 according to the user's configuration, or the result of one candidate value estimating step is determined by another candidate value. It may also be performed in a sequential structure for use in the estimation step. It is also possible to perform only one candidate value estimating step for the purpose of simplifying the method.

In the temporal candidate value estimating step (S32), motion prediction using one or more key pictures (c) reconstructed as a reference image using the candidate most similar to the corresponding position of the reconstructed image (a) as shown in FIG. 8. Find it through. As described above, the reference image used for the motion prediction may be a case of using a previous key picture, a subsequent key picture, or both before and after key pictures based on the WZ picture to be corrected and reconstructed.

Here, the reference image position most similar to the current position of the reconstructed image a is estimated by searching on the selected key picture. Such forward or reverse motion prediction or bidirectional prediction is performed to find three candidate positions having the highest similarity in each direction, and estimate the candidate position having the smallest SAD among them. Alternatively, it is also possible to approach in the direction of generating a reasonable candidate value by combining all or part of the three through a predetermined procedure without selecting one of the three found in the above process. In addition, a configuration of estimating candidate values by performing only one of forward and backward motion prediction for the purpose of reducing the amount of motion prediction may be possible. In addition, since the motion prediction of various configurations can be considered, the scope of the present invention includes all reasonable motion prediction configurations.

In the final correcting step (S33), the value of each position finally determined that an error of the reconstructed image a has occurred is corrected to the calculated candidate value.

The channel decoding error correction step S30 of FIG. 10 may be performed as shown in FIG. 13 to further increase the accuracy of the error correction through the temporal candidate value estimation described above. Referring to FIG. 13, after correcting the error occurrence position value by obtaining a correction candidate value using spatial information as in the spatial candidate value estimation step S31 of FIG. 12 through the spatial candidate value estimation and correction step S31a, The temporal candidate value prediction through the motion prediction described above is performed through the motion prediction step S31b, and the error occurrence position value is corrected again in the final correction step S31c using this value. By performing the spatial candidate value estimation and correction step (S31a), further improved error correction is possible, and the motion prediction is further performed using this value, so that more accurate candidate values can be estimated when predicting temporal candidate values. Improvements are possible.

In the embodiment according to the present invention, it is described that all encoding and decoding processes including quantization are performed in the pixel region. However, the gist of the present invention is applicable regardless of whether the encoding or decoding region is a transform region instead of the pixel region. When the present invention is applied in the transform region, as illustrated in FIG. 14, the transform unit 37 and the inverse transform unit 38 may be additionally applied.

Thus, the representation of a pixel or a corresponding position described in the description of the present invention can be considered not only as a pixel region but also as a transform coefficient in a transform region such as an integer transform, a DCT, or a wavelet transform, according to the implementation of the present invention. In the case of thinking of a transform coefficient, a transform unit and an inverse transform unit 38 shown in FIG.

As described above, a decoding error correction method and apparatus therefor in a Wyner-Ziv coding technique according to a preferred embodiment of the present invention can be made. There may be various embodiments without departing from the spirit of the invention. Therefore, the scope of the present invention should not be defined by the described embodiments, but by the claims and equivalents of the claims.

1 is a diagram illustrating a configuration of an encoder and a decoder corresponding thereto according to a conventional Wyner-Ziv coding technique.

FIG. 2 is a diagram illustrating a configuration of a channel code decoder in the Wyner-Ziv coding technique of FIG. 1.

3 is a diagram illustrating a configuration of a Wyner-Ziv encoding apparatus in a pixel region and a decoding apparatus of a distributed video encoded image using error correction including decoding error determination and correction according to the present invention.

4 is a diagram illustrating an example of a configuration of an error correction unit of the decoding apparatus of FIG. 3;

5 is a diagram illustrating an example of a configuration of a decoding error determination unit of the error correction unit of FIG. 4;

6 is a diagram illustrating an example of a configuration of a decoding error correcting unit of the error correction unit of FIG. 4;

7 is a diagram illustrating examples of specific pixels and neighboring pixels;

8 is a view showing an example of the image information used for error correction of the error correction according to the present invention,

9 is a diagram illustrating another example of a configuration of a decoding error correcting unit of the error correction unit of FIG. 4;

10 to 13 are diagrams for describing a method of decoding a distributed video encoded image using error correction according to the present invention.

14 is a diagram illustrating a configuration of a Wyner-Ziv encoding apparatus and a decoding apparatus of a distributed video encoded image using error correction including a decoding error discrimination and correction function according to the present invention.

<Explanation of symbols for the main parts of the drawings>

10: Wyner-Ziv coding device 30: decoding device

32: channel code decoding unit 33: key picture decoding unit

34: image restoration unit 35: auxiliary information generator

36: Error correction

Claims (15)

In the apparatus for decoding distributed video encoded video using error correction, A key picture decoding unit for reconstructing a key picture transmitted from the encoding device; An auxiliary information generator configured to generate auxiliary information using the key picture reconstructed by the key picture decoder; A channel code decoder for estimating a quantized value using the parity bits transmitted from the encoding apparatus and the auxiliary information; An image restoring unit for restoring a WZ picture to be decoded by using the quantized value estimated by the channel code decoding unit and the auxiliary information; Determining whether a channel code decoding error has occurred on the WZ picture using the auxiliary information and the key picture reconstructed by the key picture decoding unit, and correcting an error of the reconstructed WZ picture based on a picture similarity. Decoding device comprising an error correction. The method of claim 1, The error correction, A decoding error determination unit for determining a correction target pixel in which a channel code decoding error occurs among pixels in the restored WZ picture based on the auxiliary information and the restored key picture; And a decoding error correcting unit configured to correct the corrected pixel based on a similarity between the corrected pixel and the neighboring pixel temporally and / or spatially and determined by the decoding error determining unit. Device. The method of claim 2, The decoding error determination unit, At least one of a spatial similarity measuring unit measuring a spatial similarity between a specific pixel in the reconstructed WZ picture and a neighboring pixel, and a temporal similarity measuring unit measuring a temporal similarity between the auxiliary information and the reconstructed WZ picture; Comparing at least one of the spatial similarity and the temporal similarity with a preset threshold value, and determining that the correction target pixel exists when at least one of the spatial similarity and the temporal similarity is larger or smaller than the threshold value. And a final discriminating unit. The method of claim 3, And the spatial similarity measurer measures the spatial similarity using a difference in pixel values between the specific pixel and the neighboring pixel. The method of claim 3, And the temporal similarity measuring unit measures the temporal similarity using a difference in pixel values between the pixels corresponding to each other between the restored WZ picture and the auxiliary information. The method of claim 2, The decoding error correction unit, A spatial candidate value estimator estimating a spatial candidate value based on a spatial similarity between the correction target pixel and a neighboring pixel; and a temporal candidate value based on a temporal similarity between the correction target pixel and a corresponding pixel in the reconstructed key picture. At least one of an estimated temporal candidate value estimator; And a final correcting unit configured to correct the pixel to be corrected using at least one of the temporal candidate value and the spatial candidate value. The method of claim 6, And the spatial candidate value estimating unit estimates an intermediate value of pixel values of neighboring pixels of the correction target pixel as the spatial candidate value. The method of claim 6, The temporal candidate value estimator estimates the temporal candidate value through motion prediction for the reconstructed WZ picture by using at least one key picture reconstructed by the key picture decoder as a reference image. . In the decoding method of distributed video encoded video using error correction, (a) restoring a key picture transmitted from the encoding apparatus; (b) generating auxiliary information using the reconstructed key picture; (c) estimating a quantized value using parity bits transmitted from the encoding apparatus and the auxiliary information; (d) restoring a WZ picture to be decoded using the estimated quantized value and the auxiliary information; (e) determining whether a channel code decoding error has occurred on the WZ picture using the auxiliary information and the key picture reconstructed by the key picture decoding unit; (f) correcting an error of the reconstructed WZ picture based on a picture similarity. The method of claim 9, In step (e), At least one of measuring spatial similarity between a specific pixel in the reconstructed WZ picture and a neighboring pixel, and measuring temporal similarity between the auxiliary information and the reconstructed WZ picture; Comparing at least one of the spatial similarity and the temporal similarity with a preset threshold; And determining that there is a correction target pixel in which a channel code decoding error occurs in the reconstructed WZ picture, at least one of the spatial similarity and the temporal similarity is greater than or less than the threshold value. . The method of claim 10, The spatial similarity is measured based on a difference in pixel values between the specific pixel and the neighboring pixel. The method of claim 10, And the temporal similarity is measured based on the difference in pixel values between the corresponding pixels between the reconstructed WZ picture and the auxiliary information. The method of claim 10, Step (f), Estimating a spatial candidate value based on spatial similarity between the correction target pixel and a neighboring pixel, and estimating a temporal candidate value based on temporal similarity between the correction target pixel and a corresponding pixel in the reconstructed key picture; At least one step; And correcting the correction target pixel using at least one of the temporal candidate value and the spatial candidate value. The method of claim 13, And the spatial candidate value is estimated as an intermediate value of pixel values of neighboring pixels of the corrected pixel in the reconstructed WZ picture. The method of claim 13, And the temporal candidate value is estimated through motion prediction on the reconstructed WZ picture by using at least one key picture reconstructed by the key picture decoder as a reference image.

Images