KR20100011154A - Video encoding/decoding apparatus and pixel-based adaptive overlapped block motion compensation method and apparatus - Google Patents

Abstract

PURPOSE: A moving image encoder/decoding apparatus and the pixel-based OBMC apparatus thereof and method OBMC is proceed without the additional information of the pixel by pixel to 'the form which is applicable in the existing video compression process to the block oriented, and the block boundary unit and pixel by pixel. CONSTITUTION: The difference of the block which the difference output unit(212) is motion-compensated present, the pixel value between the movement compensation block is obtained. The standards residual pixel output unit(213) obtains the first residual pixel in which the absolute value of the pixel value most smalls and the second residual pixel in which the absolute value of the pixel value about pixels exceeding the standard value. The information recorder(214) stores the first and the second residual pixel, the big value to the first information. In case the absolute value of the inputted residual pixel as described above bigger than the value of the first information, the adaptation OBMC(overlapped block motion compensator) part(215) is adaptively proceed the nested block motion compensation and block motion compensation.

Description

Video encoding / decoding apparatus and pixel-based adaptive overlapped block motion compensation method and apparatus for same

The present invention relates to a technique for compressing and restoring video data, and more particularly, to solve a blurring artifact or an oversmoothing problem caused by overlapped block motion compensation (OBMC). A video encoding / decoding apparatus and a pixel-based adaptive superimposed block motion compensation apparatus and method therefor.

In general, most existing commercial video compression schemes and apparatuses use the Block Motion Estimation / Compensation scheme and apparatuses to effectively remove time-duplication present in natural video. This block motion estimation / compensation method is based on the assumption that all pixels included in one block basically have the same motion, and the image (s) previously compressed by each block pixel of the current image to be encoded are transmitted. ) To predict / restore. Such a block motion compensation method is based on the simple but efficient basic assumption, and has a small number of motion parameters to be coded and transmitted, thus contributing to the improvement of the compression efficiency of video data. However, blocking artifacts are caused by the incompatibility of some pixels in the block (generally the pixels at the block boundary) that do not conform to the basic model, where the block phenomenon is the block of motion estimation / reconstruction unit of each block. This means that a grid-like unnatural artificial coding error is observed at the boundary. Various prior arts have been proposed to solve such a block phenomenon, and Overlapped Block Motion Compensation is one category of such prior art.

The overlapping block motion recovery technique performs a motion recovery by weighting the reconstructed pixels at the current position due to the movement of the neighboring blocks with the reconstructed pixels of the current block when performing the motion reconstruction on each block. It is an efficient technique that can greatly reduce the motion restoration error of the block boundary by reflecting the motion to the motion restoration of the current block. However, when the motion of the adjacent block has a large difference from the current block, or when only one block of two adjacent blocks includes contour information, the motion recovery by the overlapped motion recovery method is more degraded than the existing block motion recovery. It is known to cause a blurring effect (Blurring Artifact) to show the result or blur the outline information of the block. This phenomenon is sometimes referred to as the Oversmoothing Problem. The basic assumptions of the block motion estimation method are well satisfied, so that the pixel values of the block boundary that are relatively correctly predicted are compared with those of the neighboring block that has a large error due to the overlapping block motion reconstruction. It is caused by weighted sum. In general, the difference between two adjacent blocks shows a big difference because the actual movement of most pixels of each block is different from each other, and the discontinuity may appear in the actual movement of the pixels. This is because they belong to different objects in the image, and the movement of each object is different. Therefore, the over-flat problem generally occurs in the contour portion of an object existing in the image, and since the contour information corresponds to visually more important image information, solving the over-flat problem can reduce the compression performance of the video data. It is very important to further improve and obtain a reconstructed image of better image quality.

In order to solve such over-flat problem, various techniques exist in the form of the prior art, most of which are methods of adaptively selecting and applying overlapping block motion compensation and conventional block motion compensation to motion recovery of each block. . Ji Zhongwei, Jiang Wenjun , and Zhu Weile ("Wavelet-based video coding using adaptive overlapped block motion compensation", in Proc. ICCCS'02, 29 June-1 July, 2002, vol.2, pp.1090-1093) After performing both block motion compensation and block motion compensation, the mean-squared-error of each method is compared, and the method having the smaller value is determined as the motion compensation method of the block. This technique has always achieved improved motion reconstruction performance by performing time-processing in a method with less error in motion compensation, but the average square error of each method, which is the selection criterion in the encoder, cannot be reproduced by the decoder. There is a problem in that the block should be transmitted in the form of additional information on how the motion should be restored.

While solving the problem of the increase in the compression bit rate according to the transmission of the additional information, there are various types of conventional techniques that can adaptively perform the overlap block motion compensation, first, Tien - ying Kuo and C.-C. Jay Kuo ("A hybrid BMC / OBMC motion compensation scheme", in Proc. ICIP'97, 26-29 Oct. 1997, vol. 2, pp.795-798) shows the difference image from two previously decoded images. Based on this, we propose an adaptive scheme to switch the motion compensation scheme based on the Displaced Frame Difference. Only when the difference image block having the same position as the current block to perform motion compensation has many pixels exceeding a certain threshold, the overlapped block motion compensation is performed in the decoder without additional information. This allows the selection of overlapping block motion compensation. Similarly, according to the original selection criteria for different every number of each of the technologies in the prior art, without the additional information were to be selected for each block, the motion compensation method, Lee Young - Soo ( "MPEG-4 having operating screen compensated at the time of coding Device ”, Korean Patent No. 10-280498-0000, Nov. 10, 2000, based on the high-band frequency energy value of the transmitted and decoded Discrete Cosine Transform (DCT) coefficient, Sung - hee Lee and Bong - soo Hur ("Apparatus to provide block-based motion compensation and method approximately", US Patent, PN US2004 / 0252896, Dec. 16, 2004) and Seung Hwan Kim , Dong - il Chang , Choong Woong Lee, and Sang Uk Lee ("Complexity reduction method for overlapped block motion compensation based on spatio-temporal correlation", in Proc. ISCAS'99, 30 May-2 June 1999, vol. 4, pp. 211-214), etc. And based on the motion vector difference values of the neighboring blocks and Jun Zhang ("Adaptive overlapped block matching for accurate motion compensation", US Patent, PN US2006 / 0083310, Apr. 20 2006). Overlapping block motion reconstruction is selectively applied based on the statistical standard deviation of

All of the prior art techniques described above that adaptively select motion compensation without additional information have successfully contributed to alleviating the over-flat problem by providing their own selection criteria, but they all share one important drawback. Can be. This method is based on the boundary of each block or based on each pixel in the block because each method selects a motion reconstruction method for each block and applies the selected method to perform motion reconstruction of all pixels in the block. It is impossible to restore the motion of the finer unit, and thus the ability to solve the over-flat problem may be limited.

This problem can be partially solved by the method proposed by Jiro Katto ("Overlapped motion compensation using a window function which varies in response to an input picture", US Patent, PN 5602593, Feb. 11 1997), In this method, the weight to be used in the weighted sum is adjusted according to the statistical characteristics of the input image and the motion compensation error. That is, the value of the weight to be used by the overlapped block motion compensation can be varied according to the statistical characteristics of a specific input video, a specific video of the input video, or even a specific boundary of the input video. In this case, since this method corresponds to the conventional block motion compensation method, this method can be selected to extend the existing block-based adaptive motion compensation method to select a motion restoration method more precisely and adaptively up to the block boundary unit. have. However, in this method, the statistical characteristics used as a criterion for weight adjustment include the statistical standard deviation, correlation, and motion reconstruction error of a group of pixels corresponding to the input image or part of the image. Because it means statistical expectation, etc., it takes a lot of computational amount to estimate it, and when applying this method to small areas of the image such as the boundary of a block, the uncertainty of estimation accuracy increases, which eventually limits the performance. It has the disadvantage of being visible. In addition, since such statistical parameters used for weight adjustment are not values that can be reproduced through calculation in the decoder, there is a problem that they must be transmitted to the decoder in the form of additional information.

Problems with the proposed method of Jiro Katto ("Overlapped motion compensation using a window function which varies in response to an input picture", US Patent, PN 5602593, Feb. 11 1997) are Byeong - Doo Choi , Jong-Woo Han , Chang - Su Kim , and Sung - Jae Ko ("Motion-compensated frame interpolation using bilateral motion estimation and adaptive overlapped block motion compensation", IEEE Trans. Circuits and Syst. For Video Technol., Vol. 17, pp.407-416, Apr. 2007) It is expected that the proposed scheme can be solved by modifying the proposed scheme. In this scheme, the reliability of the pixel due to the movement of the neighboring block facing each boundary of the block to be restored is similar to the actual pixel of the block to be restored currently. Based on, the higher the weight is, the higher weight is used to efficiently perform the weighted block motion compensation of the boundary unit. However, since this method is basically designed to be used in frame interpolation applications, it is impossible to reproduce reliability information, which is a criterion for weight adjustment, in the video compression application. There is a drawback that it cannot be extended by an adaptive weighted block motion compensation scheme.

Conventional techniques for adjusting the weighted block motion compensation on a pixel-by-pixel basis include Chih-lung Bruce. Lin , Ming - Chieh Lee , and The method proposed by Wei - ge Chen ("Overlapped motion compensation for object coding", US Patent, PN 5982438, Nov. 9 1999) includes a method for each pixel in a block to perform motion compensation. The super-flat problem can be completely solved by performing the overlap block motion compensation only when the pixel belongs to the same object as the pixel of the neighboring block used to obtain the weighted sum. However, this method must have an object-map in pixels and transmit it as additional information, which is generally limited by current technology to automatically generate from natural images. However, the introduction of this technique into compression coding of video information has a problem that the practicality is very low.

In order to solve the above-described problem, the present invention performs adaptive overlapping block motion compensation on a pixel basis as well as block basis, block boundary basis, and in a form applicable to existing video compression schemes without additional information in units of pixels. An object of the present invention is to provide a video encoding / decoding apparatus and a pixel-based adaptive overlapping block motion compensation apparatus and method therefor.

In order to achieve the above object, the present invention quantitatively analyzes the difference between residual pixels due to Block Motion Compensation and Overlapped Block Motion Compensation, and uses them. It is characterized by adaptively selecting block motion compensation and overlapping block motion compensation in pixel units without additional information in pixel units.

According to an aspect of the present invention to achieve the above object, a gain / loss calculation unit for obtaining the gain and loss of overlapping block motion compensation for block motion compensation; A difference calculator for calculating a difference D _N (x, y) of pixel values between a motion compensated block and a current motion compensation block using a motion vector of an adjacent block; The first residual pixel having the smallest absolute value of the pixel value is obtained for the pixels whose gain and loss exceed the reference value set based on the difference D _N (x, y), and the pixels whose gain and loss are less than or equal to the reference value. A reference residual pixel calculation unit for obtaining a second residual pixel having the largest absolute value of the pixel value with respect to; An information recording unit configured to record a larger value of the first residual pixel and the second residual pixel as a first information in a predetermined unit larger than a pixel unit; And comparing an input pixel or residual pixel with the recorded first information, and adaptively performing overlapping block motion compensation and block motion compensation when the absolute value of the input pixel or residual pixel is larger than the value of the recorded first information. If not, a pixel-based adaptive overlap block motion compensation apparatus including an adaptive overlap block motion compensation unit for performing overlap block motion compensation is provided.

When the reference residual pixel calculator adaptively performs overlapping block motion compensation and block motion compensation using a squared error, the reference value is set to 2H _N ² (x, y) XD _N ² (x, y). Where H _N (x, y) is the weight at the position (x, y) to be multiplied by the pixel block reconstructed by the motion vector of the neighboring block. The information recording unit may set the predetermined unit to, for example, one of a block boundary unit, a block unit, a part of an image, and an image unit.

The gain / loss calculating section, the difference calculating section, the reference residual pixel calculating section, and the information recording section may be provided in an encoder (or an encoding device), and the motion estimation / compensation section corresponds to the encoder. Each decoder may be provided in a decoder (or a decoding device). The encoder may transmit the first information to the decoder in the predetermined unit.

According to another aspect of the present invention for achieving the above object, (a) the gain and loss of overlapping block motion compensation for block motion compensation, the motion compensation block and the current motion compensation block using the motion vector of the adjacent block Obtaining a first residual pixel having the smallest absolute value of the pixel value for the pixels exceeding the reference value set based on the difference D _N (x, y) of the pixel values between the pixels; (b) obtaining a second residual pixel having the largest absolute value of a pixel value for pixels whose gain and loss are less than or equal to the reference value; (c) recording a larger value of the first residual pixel and the second residual pixel as first information and in a predetermined unit larger than the pixel unit; And (d) overlapping block movement when the input pixel or the received residual pixel is compared with the recorded first information, and if the absolute value of the input pixel or the received residual pixel is greater than the value of the recorded first information. There is provided a pixel-based adaptive overlapping block motion compensation method comprising adaptively performing compensation and block motion compensation, and otherwise performing overlapping block motion compensation.

When adaptively performing overlapping block motion compensation and block motion compensation using a squared error, the reference value is 2H _N ² (x, y) XD _N ² (x, y), where H _N (x, y) Is the weight at the position (x, y) to be multiplied by the pixel block reconstructed by the motion vector of the neighboring block. The predetermined unit may be, for example, one of a block boundary unit, a block unit, a part of an image, and an image unit.

Steps (a) to (c) may be performed in the encoder, and step (d) may be performed in the encoder and the decoder corresponding thereto, and the first information may be performed from the encoder to the decoder in the predetermined unit. Can be sent.

According to another aspect of the present invention to achieve the above object, block motion compensation and overlapping block motion compensation are performed based on the difference in pixel values between a motion compensated block and a current motion compensation block using motion vectors of adjacent blocks. A motion estimation / compensation unit which adaptively performs prediction to predict a predicted pixel value of each pixel of a current block of an image; A subtraction unit configured to generate a residual signal by calculating a difference value between an original pixel value of each pixel of the current block and a predicted pixel value of each pixel of the current block; A converter for converting the residual signal into frequency coefficients; A quantizer for quantizing the transformed frequency coefficients; And an encoding unit encoding the quantized frequency coefficients into a bitstream.

The motion estimation compensator may include: a gain / loss calculator for obtaining gain and loss of overlapping block motion compensation with respect to block motion compensation; A difference calculator for calculating a difference D _N (x, y) of pixel values between a motion compensated block and a current motion compensation block using a motion vector of an adjacent block; The first residual pixel having the smallest absolute value of the pixel value is obtained for the pixels whose gain and loss exceed the reference value set based on the difference D _N (x, y), and the pixels whose gain and loss are less than or equal to the reference value. A reference residual pixel calculation unit for obtaining a second residual pixel having the largest absolute value of the pixel value with respect to; An information recording unit configured to record a larger value of the first residual pixel and the second residual pixel as a first information in a predetermined unit larger than a pixel unit; And comparing the input pixel with the recorded first information, and adaptively performing overlapping block motion compensation and block motion compensation when the value of the input pixel is larger than the value of the recorded first information. In this case, the method may include an adaptive overlapping block motion compensation unit that performs overlapping block motion compensation.

According to another aspect of the present invention to achieve the above object, by comparing the input residual pixel with the set first information, overlapping block motion compensation when the value of the input residual pixel is larger than the value of the first information And an adaptive overlapping block motion compensator for adaptively performing block motion compensation, and otherwise performing overlapping block motion compensation, wherein the first information includes a motion compensated block using a motion vector of an adjacent block. An image decoding apparatus is provided based on a difference in pixel values between current motion compensation blocks.

The first information is provided from an image encoding apparatus. The image encoding apparatus includes a pixel between a motion-compensated block and a current motion compensation block whose gains and losses of overlapping block motion compensation for block motion compensation are obtained by using motion vectors of adjacent blocks. The first residual pixel having the smallest absolute value of the pixel value is obtained for the pixels exceeding the reference value set based on the difference D _N (x, y), and for the pixels whose gain and loss are less than the reference value. The second residual pixel having the largest absolute value of the pixel value may be obtained, and a larger value of the first residual pixel and the second residual pixel may be set as the first information and recorded in a predetermined unit larger than the pixel unit.

According to the present invention, it is possible to adaptively select the motion compensation scheme up to the pixel unit without the additional information in the pixel unit, thereby greatly reducing the residual signal energy of the block boundary to be encoded with very little additional information, thereby compressing the performance of the video compression apparatus. This greatly improves the efficiency of the signal and further improves the video quality when using the same bit (or the amount of information).

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. First of all, in adding reference numerals to the components of each drawing, it should be noted that the same reference numerals are used as much as possible even if displayed on different drawings. In addition, in describing the present invention, when it is determined that the detailed description of the related well-known configuration or function may obscure the gist of the present invention, the detailed description thereof will be omitted.

Since the video screen is composed of 30 frames in one second, the difference is small between one frame and neighboring frames, and thus cannot be distinguished by the human eye. Because of this, if 30 frames are sprayed in one second, the human eye perceives the frames as continuous.

As such, when the previous frame is similar to the current frame, the pixel value of the next frame can be predicted from the known pixel values constituting the previous frame (this is called interprediction).

The encoding and decoding of such video data is performed based on a motion prediction technique. The motion prediction is performed by referring to past frames or referring to both past and future frames based on the time axis. The frame referred to to encode or decode the current frame is called a reference frame. In block-based video encoding, one still image (frame) constituting a video is divided into macroblocks and subblocks constituting the macroblock, and motion is predicted in units of blocks and encoding is performed.

It is also possible to predict the next pixel using the correlation of the pixel signals in the same frame and to encode the prediction error (this is called intraprediction).

FIG. 1 is a diagram illustrating video frames constituting a video used for inter prediction.

Referring to Fig. 1, video data is composed of a series of still images. These still images are divided into GOP (Group of Picture) units. Each still image is called a frame. One GOP includes an I frame 110, a P frame 120, and a B 130 frame. The I frame 110 is a frame that is encoded by itself without using a reference frame, and the P frame 120 and the B frame 130 are frames that are encoded by performing motion estimation and compensation using the reference frame. In particular, the B frame 130 is a frame that is encoded by predicting the frame of the past and the frame of the future in the forward and reverse directions (bidirectional), respectively.

2 is a block diagram of a video encoding apparatus according to an embodiment of the present invention.

Referring to FIG. 2, the video encoding apparatus 200 according to an embodiment of the present invention may include a motion estimation / compensation unit 210, a subtraction unit 220, a transform unit 230, a quantization unit 240, and an encoding unit. 250.

The video encoding apparatus 200 may be a personal computer (PC), a notebook computer, a personal digital assistant (PDA), a portable multimedia player (PMP), or a PlayStation Portable (PSP). ), A communication device such as a communication modem for communicating with various devices or a wired / wireless communication network, a memory for storing various programs and data for encoding an image, and executing a program. Means a variety of devices including a microprocessor for operation and control.

As described above, the motion estimator / compensator 210 performs any one or a combination of intra prediction based on motion prediction or intra prediction that predicts the next pixel using correlation of pixel signals in the same frame. It can be used to predict the current block (or macroblock).

For example, the motion estimator / compensator 210 may be divided into a motion estimator (not shown) and a motion compensator (not shown). The motion estimator finds the motion prediction value of the macroblock of the current frame in the reference frame and outputs the difference of the motion as a motion vector. That is, the macroblock to be searched is searched within a predetermined search area of the reference frame to find the most similar macroblock and the degree of movement thereof is output as a motion vector. The motion compensation unit obtains a prediction macro block corresponding to the obtained motion vector from the reference frame.

As another example, the motion estimator / compensator 210 is an intra predictor that predicts a current macroblock of a current frame using a neighboring macroblock of a current macroblock in an image, and includes one or more pixel values of one or more neighboring macroblocks ( The predicted macro block is predicted by calculating a predicted pixel value of each pixel of the current macro block using the pixel value. Here, the peripheral macroblock may be one or more peripheral macroblocks that are compressed before the current macroblock and located around the current macroblock.

The subtraction unit 220 generates a residual signal by subtracting the prediction macro block from the macro block of the original image frame and calculating the difference value.

The transformer 230 converts the residual signal generated by the subtractor 220 into a frequency domain to obtain frequency coefficients. The transform unit 230 may use various transformation techniques for transforming an image signal of a time axis, such as a discrete cosine transform (DCT) or a wavelet transform, into a frequency axis. To convert the residual signal into the frequency domain. In the case of the I frame described with reference to FIG. 1, the converter 230 converts the macroblock of the original image frame into the frequency domain.

The quantization unit 240 quantizes the frequency coefficients transformed into the frequency domain by the conversion unit 230.

Subtracting the predictive macroblock from the macroblock of the original picture frame is called a residual signal, and the residual signal value is encoded to reduce the amount of data during encoding. Since an error occurs during the quantization process, the video data generated in the bitstream includes an error generated during the transformation and quantization process.

In addition, the video encoding apparatus 200 may further include an inverse quantizer 360 and an inverse transform unit 370 to obtain a reference frame.

In order to obtain a reference frame, the quantized residual signal is combined with an image predicted by the predictor 210 through an inverse quantizer 360 and an inverse transformer 270 and stored in a reference frame storage (not shown). In the case of an I frame, it is stored in the reference frame storage of the predictor 210 through the inverse quantizer 360 and the transformer 370. That is, if the original image is called A and the predicted image is called B, the converter 230 receives A-B, which is a difference between the original image and the predicted image, and performs conversion.

The encoder 250 encodes the frequency coefficients quantized by the quantizer 240 into a bitstream. An entropy encoding technique may be used as the encoding technique, but various encoding techniques may be used without being limited thereto.

3 is a block diagram of a pixel-based adaptive superimposed block motion compensation apparatus according to an embodiment of the present invention, and corresponds to the motion estimation / compensation unit 210 of FIG. 2.

As illustrated in FIG. 3, the pixel-based adaptive superimposed block motion compensation apparatus 210 according to an embodiment of the present invention may include a gain / loss calculator for obtaining gain and loss of superimposed block motion compensation for block motion compensation ( 211); A difference calculator 212 for calculating a difference D _N (x, y) between pixel values of a motion compensated block and a current motion compensation block using a motion vector of an adjacent block; The first residual pixel having the smallest absolute value of the pixel value is obtained for the pixels whose gain and loss exceed the reference value set based on the difference D _N (x, y), and the pixels whose gain and loss are less than or equal to the reference value. A reference residual pixel calculator 213 for obtaining a second residual pixel having the largest absolute value of the pixel value with respect to the? An information recording unit (214) for recording a larger value of the first residual pixel and the second residual pixel as first information in a predetermined unit larger than a pixel unit; And comparing the input pixel or residual pixel with the recorded first information, and adaptively performing overlapping block motion compensation and block motion compensation when the value of the input pixel or residual pixel is larger than the value of the recorded first information. If not, it includes an adaptive overlap block motion compensation unit 215 for performing overlap block motion compensation.

4 is a block diagram of a video decoding apparatus according to an embodiment of the present invention.

As shown in FIG. 4, the video decoding apparatus 400 according to an embodiment of the present invention predicts a current block of an image using one or more neighboring blocks around the current block and decodes the image. The apparatus includes a decoder 410, an inverse quantizer 420, an inverse transform unit 430, an adder 440, and a motion estimation / compensation unit 450.

The video decoding apparatus 400, like the video encoding apparatus 200 described above with reference to FIG. 2, may be a personal computer (PC), a notebook computer, a personal digital assistant (PDA), a portable multimedia player ( PMP: Portable Multimedia Player (PSP), PlayStation Portable (PSP: PlayStation Portable), Mobile Communication Terminal (Mobile Communication Terminal) and the like, communication devices, such as a communication modem for performing communication with various devices or wired and wireless communication network, decoding the image Means a variety of devices including a memory for storing various programs and data, a microprocessor for executing and controlling the program by executing.

The decoder 410 extracts the quantized frequency coefficients by decoding the bitstream. That is, the decoder 410 extracts the quantized frequency coefficients including pixel information of the current block of the image by decoding the bitstream which is the image encoded by the image encoding apparatus 200.

The dequantizer 420 de-quantizes the frequency coefficients extracted from the bitstream by the decoder 410.

The inverse transformer 430 generates a residual signal by inversely transforming the frequency coefficients inversely quantized by the inverse quantizer 420 into the time domain.

The adder 430 reconstructs the original pixel value of the current block by adding the residual signal inversely transformed by the inverse transform unit 430 and the predicted pixel value of each pixel of the current block predicted by the motion estimation / compensator 450. .

FIG. 5 is a block diagram of a pixel-based adaptive superimposed block motion compensation apparatus according to another embodiment of the present invention, and corresponds to the motion estimation / compensation unit 450 of FIG. 4.

As shown in FIG. 5, the pixel-based adaptive superimposed block motion compensation apparatus 450 according to another embodiment of the present invention compares the input residual pixel with the set first information, and determines the absolute value of the input residual pixel. If it is larger than the value of the first information, an adaptive overlapping block motion compensation unit 455 for adaptively performing overlapping block motion compensation and block motion compensation, and otherwise performing overlapping block motion compensation.

The first information is generated and provided from the motion estimation / compensation unit 210 of the video encoding apparatus 200 of FIG. 2. As described above, the motion compensated block and the current motion compensation block using the motion vectors of adjacent blocks. It is set based on the difference in pixel values. The process of generating the first information is the same as the embodiment described with reference to FIGS. 2-3 and 6.

FIG. 6 is a flowchart of a pixel-based adaptive superimposed block motion compensation method according to an embodiment of the present invention, and the motion estimation / compensation unit 210 of FIG. 2, that is, the pixel-based adaptive superimposed block motion compensation apparatus 210 of FIG. 3. The description will apply to.

First, the gain / loss calculator 211 calculates gain and loss of the overlapped block motion compensation for block motion compensation, and the difference calculator 212 uses the motion vector of the adjacent block to compensate for the motion and the current motion. After calculating the difference D _N (x, y) of the pixel values between the compensation blocks, the gain and loss calculated by the reference residual pixel calculator 213 are set based on the calculated difference D _N (x, y). The first residual pixel having the smallest absolute value of the pixel value is calculated for the pixels exceeding the reference value (S610).

Subsequently, the reference residual pixel calculator 213 calculates a second residual pixel having the largest absolute value of the pixel value for pixels whose gain and loss are equal to or less than the reference value (S620).

Subsequently, the information recording unit 214 uses a larger value of the first residual pixel and the second residual pixel as the first information, and a predetermined unit larger than the pixel unit, for example, a block boundary unit, a block unit, and a part of an image. In operation S630, one unit and one image unit are recorded.

Finally, in the adaptive superimposition motion compensator 215, the absolute value of the input pixel or the received residual pixel is compared with the recorded first information, so that the absolute value of the input pixel or the received residual pixel is recorded. If it is larger than the value of the first information, the overlapped block motion compensation and the block motion compensation are adaptively performed. Otherwise, the overlapped block motion compensation is performed (S640).

As a pixel-based adaptive superimposed block motion compensation method according to another embodiment of the present invention, step S640 of FIG. 6 is the motion estimation / compensation unit 450 of FIG. 4, that is, the pixel-based adaptive superimposed block motion compensation apparatus 450 of FIG. 5. In the adaptive overlapping motion compensator 455 of FIG. In this case, the first information is generated from the motion estimation / compensation unit 210 of the video encoding apparatus 200 of FIG. 2 and provided to the video decoding apparatus 400 of FIG. 4.

Next, a specific example according to the present invention will be described for a more detailed understanding of the present invention described above.

For detailed description of the present invention, first, an overlapping block motion compensation technique of H.263 ("Video coding for low bit rate communication", Draft, ITU-T Recommendation H.263, Sept. 1997) will be described.

Overlapping block motion compensation in H.263 is performed by the following equation in the unit of 8x8 block which is a unit of discrete cosine transform of residual signal.

[Equation 1]

는 8x8 블록 내 화소의 위치를 표현하는 것으로 블록의 최 좌상 위치를 (0,0)으로 하여 각기 0에서 7까지의 값을 가지게 된다. 또한

는 현재 복원 대상인 8x8 블록의 움직임 보상 화소 값을 의미하는 것으로, 현재 블록을 위해 전송되어 온 움직임 벡터를

라 하면, 다음의 [식2]를 통해 구해진다.Here, means the motion compensation pixel value to be generated by the overlap block motion compensation of H.263, and coordinates

Denotes the position of the pixel in the 8x8 block, and has a value from 0 to 7, with the leftmost position of the block as (0,0). Also

Denotes the motion compensation pixel value of the 8x8 block to be restored currently, and indicates the motion vector transmitted for the current block.

In this case, the following equation is obtained.

[Equation 2]

는 이전 복원된 영상의

위치에서의 화소 값을 의미한다.here

Of the previously restored image

It means the pixel value at the position.

은 현재 복원 대상인 블록에 인접한 인접 경계 블록 (Neighboring Block)을 지칭하는 것으로,

,

,

, 혹은

의 값을 가지게 되며, 이는 각각 현재 블록의 위쪽 (Top), 아래쪽 (Bottom), 왼쪽 (Left), 그리고 오른쪽 (Right) 경계에 인접한 블록 인덱스를 의미한다. 따라서

일 때,

는

와 같이 구해지고, 여기서

는 현재 블록 위쪽에 인접한 블록을 위해 전송되어 온 움직임 벡터를 뜻한다. 이와 유사하게,

이 다른 값을 가지는 경우에도 인접 블록의 움직임 벡터와 이전 복원 영상을 이용하여

를 구할 수 있다. 마지막으로, [식1]에서

와

는 현재 블록의 움직임 벡터로 복원 된 화소 블록

과 주변 블록의 움직임 벡터로 복원 된 화소 블록

에 곱해 질

위치에서의 가중치를 의미하는데, 각 위치에서의 가중치 행렬을 도 7에 나타내었다.In [Equation 1],

Refers to a neighboring block adjacent to the block currently being restored.

,

,

, or

It has the value of, which means the block index adjacent to the Top, Bottom, Left, and Right boundaries of the current block, respectively. therefore

when,

Is

Is obtained as

Denotes a motion vector transmitted for a block adjacent to the current block. Similarly,

Even with this different value, the motion vector of the adjacent block and the previous reconstructed image are used.

Can be obtained. Finally, in [Equation 1]

Wow

Pixel block reconstructed with the motion vector of the current block

Pixel blocks reconstructed with the motion vectors of the neighboring blocks

To be multiplied by

Meaning the weight at the position, the weight matrix at each position is shown in FIG.

As described above, the H.263 overlapping block motion compensation scheme performs motion compensation through [Equation 1] every 8x8 block by using a statistically optimized fixed weight matrix, so that the block compared with the conventional block motion compensation Significantly reduces the effect. However, it can be seen that some blocks of the reconstructed image generate a residual signal that is significantly degraded as a result of the conventional block motion compensation, and the above-described conventional techniques have their own original criteria for distinguishing some of these blocks. It corresponds to the provision. Also, among the overlapping block motion compensated blocks that show better results than the block motion compensation result, it can be observed that not all adjacent boundary blocks have a positive effect on the overlapping block motion compensation result, which is higher. In order to improve performance, it is possible to provide a reason for adaptively applying overlapping block motion compensation in a block boundary unit rather than a block unit, or more precisely, a pixel unit. In order to solve the above-mentioned problem, when the block motion compensation and the overlapping block motion compensation are performed, how the residual signal varies for each adjacent boundary of one block will be described.

를 재 정의한다.First, in order to observe the residual signal change of each block boundary by each motion compensation method, the weight of each boundary to be applied to the current restoration target block as follows.

Redefine.

[Equation 3]

은 앞서와 마찬가지로

,

,

, 혹은

중 하나의 값을 가진다.here,

As before

,

,

, or

Has one of the values.

라 정의하고, 현재 블록의 움직임 벡터를 이용해 블록 움직임 복원 된 화소를

, 그리고 인접 블록

의 움직임 벡터를 이용해 블록 움직임 복원된 화소를

라 하면, 블록 움직임 복원을 적용한 경우의 잔여 신호

와 중첩 블록 움직임 보상을 적용한 경우의 잔여신호

는 각기 다음과 같이 표현될 수 있다.Pixel value of the current image to be encoded

The pixel whose block motion is reconstructed using the motion vector of the current block

, And adjacent blocks

Block motion reconstructed pixel using

In other words, the residual signal when block motion recovery is applied

And residual signal when overlapping block motion compensation is applied

Can be expressed as

[Equation 4]

[Equation 5]

는 인접 블록

의 움직임 벡터를 통한 현재 블록 위치에서의 움직임 복원이 현재 블록의 움직임 벡터를 이용한 움직임 보상과 얼마나 차이가 나는지를 나타내는 차이(Difference) 측도에 해당한다.Here, defined as the following [Equation 6]

Adjacent blocks

This corresponds to a difference measure indicating how far the motion reconstruction from the current block position through the motion vector of Δ is different from the motion compensation using the motion vector of the current block.

[Equation 6]

위치에서의 잔여 신호를

라 하고,

인 경우를 생각해 보자. 평균 제곱 오차를 기준으로 움직임 보상 방식을 선택했기 때문에, 이는

임을 의미한다. 따라서 [식 5]를 통하여 다음을 알 수 있다.Now, suppose that the encoder, that is, the video encoding apparatus 200 as shown in FIG. 2, adaptively applies block motion compensation and overlapping block motion compensation for each pixel through a predetermined reference to a specific adjacent boundary block of one block. . Although this predetermined criterion may be in various forms, the present invention uses a squared error for the specificity and convenience of description. However, this does not mean that the present invention can adaptively select the pixel-specific motion compensation limited to the square error, and can be applied to various other types of measures through a similar derivation process as described later. First, received and decoded

The residual signal at the position

,

Consider the case. Since we chose motion compensation based on the mean squared error,

Means. Therefore, the following can be seen through [Equation 5].

[Equation 7]

인 경우, 즉

인 경우에도, [식 5]를 통하여 다음을 알 수 있다.Likewise,

If

Even in the following equation, the following can be seen.

[Equation 8]

라 놓으면, 중첩 블록 움직임 보상을 적용한 경우와 기존의 블록 움직임 보상을 적용한 경우에, 수신된 잔여 화소

을 이용하여 계산될 수 있는 결정 인자

의 범위는 도 8과와 같다. 도 8을 보면, 부호기 즉, 도 2와 같은 동영상 부호화 장치(200)에서 제곱 오차를 기준으로 화소 단위 블록 움직임 보상과 중첩 블록 움직임 보상을 수행하는 경우, 전송된 잔여 화소만으로는 부호기의 판단을 재현할 수 없는 영역이 존재한다는 것을 알 수 있다. 하지만, H.263과 같이 중첩 블록 움직임 보상을 수행하는 경우, 도 8에서 판단이 가능한 영역에서 얻을 수 있는 중첩 블록 움직임 보상의 이득과 손실을 다음의 [식 9] 및 [식 10]과 같이 생각해 보면, 수신된 잔여 신호로 부호기의 적응적 선택을 판단할 수 있는 영역은 큰 이득이나 큰 손실이 발생 하는 경우임을 알 수 있다.In equations (7) and (8),

In other words, when the overlapped block motion compensation is applied and the existing block motion compensation is applied, the received residual pixel is received.

Determinants that can be calculated using

The range of is as shown in FIG. Referring to FIG. 8, when the encoder, that is, the video encoding apparatus 200 as illustrated in FIG. 2 performs block-by-pixel block motion compensation and superimposed block motion compensation on the basis of the squared error, the judgment of the encoder may be reproduced using only the transmitted residual pixels. It can be seen that there is an unknown area. However, in the case of performing the overlapped block motion compensation as shown in H.263, the gain and loss of the overlapped block motion compensation obtained in the region that can be determined in FIG. 8 are considered as in Equations 9 and 10 below. As can be seen, the area where the adaptive selection of the encoder can be determined based on the received residual signal is a case where a large gain or a large loss occurs.

[Equation 9]

[Equation 10]

Therefore, the area where the over-flat problem occurs mostly corresponds to the contour portion or the texture area of the object in the image. Considering that the region generally generates a large loss of overlapping block motion compensation, the following specific embodiments are described. It is possible to solve the over-flat problem using [Equation 9] and [Equation 10] through.

The following is a specific example of the method of the present invention, which can adaptively select a method of pixel-by-pixel motion compensation using the above-described residual signal analysis, that is, Equation 9 and Equation 10. The adaptive overlapping block motion compensator 210 and the method according to the present invention of FIG. 6 will be described.

The following specific embodiments described below are merely examples of adaptively selecting the overlapping block motion compensation and block motion compensation scheme in pixel units without additional information in pixel units. It is not limited to the embodiment described later.

Specific Example 1

에서 아래의 단계 1 내지 단계 4를 수행하고, 수행 된 결과에

에서 아래의 단계 1 내지 단계 4를 수행한다.

Follow steps 1 to 4 below and on the result

Perform steps 1 to 4 below.

를 구한다. 즉, 도 6의 단계 S610과 같이, 이득/손실 산출부(211)에서 블록 움직임 보상에 대한 중첩 블록 움직임 보상의 이득 G(x,y)과 손실 L(x,y)을 산출하고, 차이 산출부(212)에서 [식 6]을 이용하여 인접한 블록의 움직임 벡터를 사용하여 움직임 보상된 블록과 현재 움직임 보상 블록 간의 화소 값의 차이 D_N(x,y)를 산출한 후, 기준 잔여 화소 산출부(213)에서 상기 산출된 이득 G(x,y)과 손실 L(x,y)이 상기 산출된 차이 D_N(x,y)를 기반으로 설정된 기준값 2H_N ²(x,y) x D_N ²(x,y)를 초과하는 화소들에 대하여(즉, [식 9]와 [식 10]을 만족하는 모든 화소에 대하여) 그 화소값의 절대치가 가장 작은 제 1 잔여화소 |

|를 산출한다.Step 1: First residual pixel having the smallest absolute value of pixel value for all pixels satisfying [Equation 9] and [Equation 10]

Obtain That is, as in step S610 of FIG. 6, the gain / loss calculator 211 calculates the gain G (x, y) and the loss L (x, y) of the overlapped block motion compensation for the block motion compensation, and calculates the difference. After calculating the difference D _N (x, y) of the pixel value between the motion compensated block and the current motion compensation block using the motion vector of the adjacent block by using Equation 6, the reference residual pixel is calculated using Equation 6 below. In the unit 213, the calculated gain G (x, y) and loss L (x, y) are set based on the calculated difference D _N (x, y) 2H _N ² (x, y) x D _For pixels exceeding _N ² (x, y) (ie, for all pixels satisfying [Formula 9] and [Formula 10]), the first residual pixel having the smallest absolute value of the pixel value |

Yield |

를 구한다. 즉, 도 6의 단계 S620과 같이, 기준 잔여 화소 산출부(213)에서 상기 산출된 이득 G(x,y)과 손실 L(x,y)이 상기 기준값 2H_N ²(x,y) x D_N ²(x,y) 이하인 화소들에 대하여(즉, [식 9]와 [식 10]을 만족하지 않는 나머지 화소들에 대하여) 그 화소값의 절대치가 가장 큰 제 2 잔여화소를 산출한다.Step 2: The second residual pixel having the largest absolute value of the pixel value for the remaining pixels that do not satisfy the expressions [9] and [10].

Obtain That is, as in step S620 of FIG. 6, the gain G (x, y) and the loss L (x, y) calculated by the reference residual pixel calculator 213 are the reference value 2H _N ² (x, y) x D. _For pixels that are equal to or less than _N ² (x, y) (ie, for the remaining pixels that do not satisfy [Equation 9] and [Equation 10]), a second residual pixel having the largest absolute value of the pixel value is calculated.

|및 상기 제 2 잔여화소 |

| 중 더 큰 값을 화소 단위보다 큰 일정 단위로 기록한다. 즉, 6의 단계 S630과 같이, 정보 기록부(214)에서 상기 제 1 잔여화소 |

| 및 상기 제 2 잔여화소 |

| 중 더 큰 값을 제 1 정보로 하여 화소 단위보다 큰 일정 단위, 예를 들어, 블록 경계 단위, 블록 단위, 영상의 일부 단위 및 영상 단위 중 하나의 단위로 기록한다.Step 3: The first residual pixel obtained in step 1 and step 2 |

| And the second remaining pixel |

| The larger value is recorded in a certain unit larger than the pixel unit. That is, as in step S630 of 6, the information recording unit 214 makes the first remaining pixel |

| And the second residual pixel |

| The larger value is used as the first information and is recorded in one unit larger than the pixel unit, for example, a block boundary unit, a block unit, a part of an image, and an image unit.

의 절대치를 상기 기록된 제 1 정보와 비교하여, 수신된 잔여 화소

의 절대값이 상기 기록된 제 1 정보의 값보다 더 큰 경우 중첩 블록 움직임 보상과 블록 움직임 보상을 적응적으로 수행하고, 그렇지 않은 경우에 대해서는 중첩 블록 움직임 보상을 실시한다. 즉, 도 6의 단계 S640과 같이, 적응 중첩 움직임 보상부(215,455)에서 수신된 잔여화소

의 절대치를 상기 기록된 제 1 정보와 비교하여, 상기 수신된 잔여화소

의 절대값이 상기 기록된 제 1 정보의 값보다 큰 경우 중첩 블록 움직임 보상과 블록 움직임 보상을 적응적으로 수행하고, 그렇지 않은 경우에 대해서는 중첩 블록 움직임 보상을 실시한다.Step 4: Finally, the coded or decoded terminal receives the remaining pixels

The received residual pixel by comparing the absolute value of

If the absolute value of is larger than the value of the recorded first information, the overlapping block motion compensation and the block motion compensation are adaptively performed. Otherwise, the overlapping block motion compensation is performed. That is, as in step S640 of FIG. 6, the residual pixels received by the adaptive overlapping motion compensators 215 and 455.

The received residual pixel by comparing the absolute value of

If the absolute value of is larger than the value of the recorded first information, the overlapped block motion compensation and the block motion compensation are adaptively performed. Otherwise, the overlapped block motion compensation is performed.

The above description is merely illustrative of the technical idea of the present invention, and those skilled in the art to which the present invention pertains may make various modifications and changes without departing from the essential characteristics of the present invention. Therefore, the embodiments disclosed in the present invention are not intended to limit the technical idea of the present invention but to describe the present invention, and the scope of the technical idea of the present invention is not limited by these embodiments. The scope of protection of the present invention should be interpreted by the following claims, and all technical ideas within the scope equivalent thereto should be construed as being included in the scope of the present invention.

As described above, the present invention is applied to a video encoding and decoding technology field, and enables a motion compensation scheme to be adaptively selected up to a pixel level without additional information of a pixel unit, thereby requiring a block boundary portion to be encoded with very little additional information. It is a very useful invention that greatly reduces the residual signal energy, thereby greatly improving the compression performance of the video compression apparatus, and furthermore, when using the same bit (or information amount), an improved video quality can be obtained.

1 is a view showing video frames constituting a video, used for inter prediction;

2 is a block diagram of a video encoding apparatus according to an embodiment of the present invention;

3 is a block diagram of a pixel-based adaptive overlap block motion compensation apparatus according to an embodiment of the present invention;

4 is a block diagram of a video decoding apparatus according to an embodiment of the present invention;

5 is a block diagram of a pixel-based adaptive overlap block motion compensation apparatus according to another embodiment of the present invention;

6 is a flowchart of a pixel-based adaptive overlap block motion compensation method according to an embodiment of the present invention;

7 is a diagram illustrating a weight matrix of H.263 overlapping block motion recovery according to an embodiment of the present invention;

FIG. 8 is a diagram illustrating a range of determinants calculated from received residual histories according to an embodiment of the present invention.

<Description of Symbols for Main Parts of Drawings>

210: motion estimation / compensation unit, pixel-based adaptive overlap block motion compensation device

211: gain / loss calculator 212: difference calculator

213: reference pixel calculating unit 214: information recording unit

215: adaptive overlapping block motion compensator

220: subtraction unit 230: conversion unit

240: quantizer 250: encoder

260: inverse quantization unit 270: inverse transform unit

410: decoder 420: inverse quantization unit

430: inverse transform unit 440: adder

450: motion estimation / compensation unit, pixel-based adaptive nesting block motion compensation device

455: adaptive overlapping block motion compensator

Claims (15)

A gain / loss calculator for calculating gains and losses of overlapping block motion compensation for block motion compensation; A difference calculator for calculating a difference D _N (x, y) of pixel values between a motion compensated block and a current motion compensation block using a motion vector of an adjacent block; The first residual pixel having the smallest absolute value of the pixel value is obtained for the pixels whose gain and loss exceed the reference value set based on the difference D _N (x, y), and the pixels whose gain and loss are less than or equal to the reference value. A reference residual pixel calculation unit for obtaining a second residual pixel having the largest absolute value of the pixel value with respect to; An information recording unit configured to record a larger value of the first residual pixel and the second residual pixel as a first information in a predetermined unit larger than a pixel unit; And Comparing the input residual pixel with the recorded first information, and adaptively performing overlapping block motion compensation and block motion compensation when the absolute value of the input residual pixel is larger than the value of the recorded first information; If not, an adaptive overlapping block motion compensator for performing overlapping block motion compensation Pixel-based adaptive superimposed block motion compensation apparatus comprising a. The method of claim 1, The reference residual pixel calculator sets the reference value to 2H _N ² (x, y) x D _N ² (x, y) when adaptively performing overlapping block motion compensation and block motion compensation using a squared error. And wherein H _N (x, y) is a weight at a position (x, y) to be multiplied by the pixel block reconstructed by the motion vector of the neighboring block. The method of claim 1, And the information recording unit sets the predetermined unit to one of a block boundary unit, a block unit, a partial unit of an image, and an image unit. The method of claim 1, The gain / loss calculator, the difference calculator, the reference residual pixel calculator, and the information recorder are provided in an encoder, and the motion estimator / compensator is provided in the encoder and a decoder corresponding thereto. Pixel-based Adaptive Overlapping Block Motion Compensation Device. The method of claim 4, wherein And the encoder transmits the first information to the decoder on a predetermined basis. (a) Gain and loss of overlapping block motion compensation for block motion compensation is based on the difference D _N (x, y) between the pixel value of the motion compensated block and the current motion compensation block using the motion vector of the adjacent block. Obtaining a first residual pixel having the smallest absolute value of the pixel value for the pixels exceeding the set reference value; (b) obtaining a second residual pixel having the largest absolute value of a pixel value for pixels whose gain and loss are less than or equal to the reference value; (c) recording a larger value of the first residual pixel and the second residual pixel as first information and in a predetermined unit larger than the pixel unit; And (d) adaptively superimpose block motion compensation and block motion compensation when the absolute value of the received residual pixel is greater than the value of the recorded first information by comparing the absolute value of the received residual pixel with the recorded first information. If not, perform the overlap block motion compensation if not Pixel-based adaptive overlapping block motion compensation method comprising a. The method of claim 6, When adaptively performing overlapping block motion compensation and block motion compensation using a squared error, the reference value is 2H _N ² (x, y) XD _N ² (x, y), where H _N (x, y) Is a weight at the position (x, y) to be multiplied by the pixel block reconstructed by the motion vector of the neighboring block. The method of claim 6, And the predetermined unit is one of a block boundary unit, a block unit, a partial unit of an image, and an image unit. The method of claim 6, The steps (a) to (c) are performed in the encoder, and the step (d) is performed in the encoder and the corresponding decoder, respectively. The method of claim 9, And the first information is transmitted from the encoder to the decoder on a predetermined basis. Based on the difference in pixel values between the motion compensated block and the current motion compensation block using the motion vectors of the adjacent blocks, the block motion compensation and the overlapping block motion compensation are adaptively performed to predict the pixels of each pixel of the current block of the image. A motion estimator / compensator for predicting a value; A subtraction unit configured to generate a residual signal by calculating a difference value between an original pixel value of each pixel of the current block and a predicted pixel value of each pixel of the current block; A converter for converting the residual signal into frequency coefficients; A quantizer for quantizing the transformed frequency coefficients; And An encoder that encodes the quantized frequency coefficients into a bitstream Video encoding apparatus comprising a. The method of claim 11, The motion estimation compensator, A gain / loss calculator for calculating gains and losses of overlapping block motion compensation for block motion compensation; A difference calculator for calculating a difference D _N (x, y) of pixel values between a motion compensated block and a current motion compensation block using a motion vector of an adjacent block; The first residual pixel having the smallest absolute value of the pixel value is obtained for the pixels whose gain and loss exceed the reference value set based on the difference D _N (x, y), and the pixels whose gain and loss are less than or equal to the reference value. A reference residual pixel calculation unit for obtaining a second residual pixel having the largest absolute value of the pixel value with respect to; An information recording unit configured to record a larger value of the first residual pixel and the second residual pixel as a first information in a predetermined unit larger than a pixel unit; And comparing the input pixel with the recorded first information, and adaptively performing overlapping block motion compensation and block motion compensation when the absolute value of the input pixel is larger than the value of the recorded first information. In this case, the video encoding apparatus is configured to include an adaptive overlapping block motion compensator for performing overlapping block motion compensation. When the absolute value of the input residual pixel is larger than the value of the first information, the overlapping block motion compensation and the block motion compensation are adaptively performed, compared with the set first information. An adaptive overlapping block motion compensator for performing overlapping block motion compensation is provided. And the first information is set based on a difference in pixel values between a motion compensated block and a current motion compensation block using a motion vector of an adjacent block. The method of claim 13, The gain and loss of the overlapped block motion compensation for the block motion compensation are based on the reference value set based on the difference D _N (x, y) of the pixel value between the motion compensated block and the current motion compensation block using the motion vector of the adjacent block. The first residual pixel having the smallest absolute value of the pixel value is obtained for the excess pixels, and the second residual pixel having the largest absolute value of the pixel value is obtained for the pixels whose gain and loss are less than or equal to the reference value. And a larger value of the residual pixel and the second residual pixel is set as the first information. The method of claim 14, And the first information is recorded in one of a block boundary unit, a block unit, a partial unit of an image, and an image unit in the video encoding apparatus and transmitted to the video decoding apparatus.

Images