WO2008111744A1

WO2008111744A1 - Method and apparatus for encoding and decoding image in pixel domain

Info

Publication number: WO2008111744A1
Application number: PCT/KR2008/001115
Authority: WO
Inventors: Hae-Kwang Kim
Original assignee: Industry-Academia Cooperation Group Of Sejong University
Priority date: 2007-03-13
Filing date: 2008-02-26
Publication date: 2008-09-18

Abstract

Provided are a method and apparatus for encoding and decoding an image. The method of encoding an image includes: generating a first residual block of a current block using intra prediction or inter prediction; loss encoding the first residual block in frequency domain; decoding the encoded first residual block and restoring a second residual block in pixel domain; and entropy coding the restored second residual block. Therefore, an image can be easily restored based on addition and reduction of pixel domain without complex inversion operations in the frequency domain in decoding the encoded image.

Description

METHOD AND APPARATUS FOR ENCODING AND DECODING IMAGE IN PIXEL DOMAIN

TECHNICAL FIELD

The present invention relates to a method and apparatus for encoding and decoding an image, and more particularly, to a method and apparatus for encoding and decoding an image in pixel domain so as to not perform a complex operation in frequency domain while decoding an image.

BACKGROUND ART

In image data compression such as MPEG-1 , MPEG-2, and MPEG-4 H.264/MPEG-4 Advanced Video coding (AVC), one picture is divided into a plurality of blocks to encode an image. Then, each of the blocks are encoded using inter prediction or intra prediction.

The inter prediction is to remove temporal redundancy between pictures so as to compress an image. Examples of inter prediction include motion estimation encoding. The motion estimation encoding estimates motion of each block included in a current picture using one ore more reference pictures. In addition, the motion estimation encoding searches for a reference block that is the most similar to the current block in a predetermined searching range of one or more reference pictures using a predetermined evaluation function so as to estimate the current block based on the search result.

The intra prediction uses pixel values included in a previously encoded region that is spatially adjacent to a block to be encoded, instead of referring the reference pictures for encoding the current block. The intra prediction is performed in a predetermined direction based on the pixel values included in the previously encoded region so as to estimate the current block.

FIG. 1 is a block diagram of a conventional apparatus for encoding an image.

Referring to FIG. 1 , the conventional apparatus for encoding an image 100 includes a motion estimation unit 102, a motion compensation unit 104, an intra estimation unit 106, a transformer 108, a quantization unit 110, an entropy coding unit 112, a dequantization unit 114, an inverse transformer 116, a filter 118, and a frame memory 120.

The motion estimation unit 102 refers to at least one reference picture stored in the frame memory 120 and estimates motion of the current block. The block having the smallest Sum of Absolute Difference (SAD) with the current block is searched from at least one reference picture. According to the search result, motion vector of the current block is generated. The motion compensation unit 104 compensates the current block based on the motion estimation result of the motion estimation unit 102. The motion estimation result of the motion estimation unit 102 is used to generate a prediction block of the current block.

The intra estimation unit 106 intra estimates the current block using pixels included in the previously encoded region that is adjacent to the current block. The prediction block of the current block is generated by performing intra prediction in a predetermined direction based on pixel values of the pixels included in the previously encoded region that is adjacent to the current block.

The current block is subtracted by the prediction block generated by the motion compensation unit 104 or the intra estimation unit 106 so as to generate a residual block DN. The generated residual block DN is discrete cosine transformed (DCT) in the transformer 108. As the result of the DCT, DCT coefficients are generated and are quantized in the quantization unit 110.

The entropy coding unit 112 entropy codes the DCT coefficients quantized in the quantization unit 110 and generates bitstream including data for the current block.

The quantized DCT coefficients are dequantized and inverse discrete cosine transformed (IDCT) respectively in the dequantization unit 114 and the inverse transformer 116 and are restored as a residual block D'n. The restored residual block D'n is different from the residual block DN input to the transformer 108. While quantizing the DCT coefficient in the quantization unit 110, loss causes and the DCT coefficients from which the loss causes are dequantized and IDCT so that the residual block D'n containing the loss is restored. The restored residual block D'n is added to the prediction block to restore the current block. The filter 118 stores the restored current block in the frame memory 120 by deblocking filtering the restored current block. The restored current block is also different from the current block prior to be encoded. This is because the residual block including the loss is added to the estimation block to restore the current block.

The conventional apparatus for encoding an image 100 discrete cosine transforms an image to transform to the frequency domain and then quantized to loss code the current block. Encoding is performed after transforming to the frequency domain since eyes of human beings are insensitive to high frequency components and sensitive to low frequency components. Also, loss encoding is performed through quantization since compression rate can be increased by loss encoding, instead of lossless encoding.

However, when an image is encoded or decoded using the conventional art, a number of operations should be carried out for encoding and decoding an image in the frequency domain using the DCT. Thus, encoding and decoding speed are lowered and hardware resources of apparatuses for encoding and decoding used in image processing should be sufficient.

DETAILED DESCRIPTION OF THE INVENTION

TECHNICAL PROBLEM

The present invention provides a method and apparatus for encoding and decoding an image capable of loss encoding an image in frequency domain and rapidly decoding an image, and a computer readable recording medium having embodied thereon a computer program for executing the method.

TECHNICAL SOLUTION

According to an aspect of the present invention, there is provided a method of encoding an image including: generating a first residual block of a current block using intra prediction or inter prediction; loss encoding the first residual block in frequency domain; decoding the encoded first residual block and restoring a second residual block in pixel domain; and entropy coding the restored second residual block.

The entropy coding may include: predicting each of residual values included in the restored second residual block from another residual value adjacent to a vertical direction to the each of residual values or another residual value adjacent to a horizontal direction to the each of residual values and generating a third residual block based on the result of the prediction; and entropy coding the generated third residual block.

The entropy coding may include: comparing the residual values of the first residual block and the second residual block and generating a fourth residual block only including the residual values having the small absolute value; predicting each of residual values of the fourth residual block from another residual value adjacent to a vertical direction to the each of residual values of the fourth residual block or another residual value adjacent to a horizontal direction to the each of residual values and generating a third residual block based on the result of the prediction of the residual values included in the fourth residual block; and entropy coding the generated third residual block.

According to another aspect of the present invention, there is provided a method of decoding an image including: entropy decoding data for a residual block of a current block and restoring the residual block; and restoring the current block based on the restored residual block, wherein the data for the residual block is for a second block restored by decoding a first residual block loss encoded in frequency domain, and wherein the first residual block of the current block is generated by performing intra prediction or inter prediction while encoding the current block.

According to another aspect of the present invention, there is provided an apparatus for encoding an image including: a prediction unit which generates an prediction block of a current block using intra prediction or inter prediction; a subtracting unit which generates a first residual block by subtracting the generated prediction block from the current block; a residual block generator which loss encodes the first residual block in frequency domain, decodes the encoded first residual block, and restores a second residual block of pixel domain; and a lossless encoding unit which entropy encodes the restored second residual block.

According to another aspect of the present invention, there is provided an apparatus for decoding an image including: a lossless decoding unit which entropy decodes data for a residual block of a current block and restores the residual block; and a restoring unit which restores the current block based on the restored residual block, wherein the data for the residual block is for a second block restored by decoding a first residual block loss encoded in frequency domain, and wherein the first residual block of the current block is generated by performing intra prediction or inter prediction while encoding the current block.

According to another aspect of the present invention, there is provided a computer readable recording medium having embodied thereon a computer program for executing the method described above.

ADVANTAGEOUS EFFECTS According to the present invention, an image is loss encoded in frequency domain so that compression rate of image encoding can be increased. Also, an image can be easily restored based on addition and reduction of pixel domain without complex inversion operations in the frequency domain in decoding the image so that image-decoding speed is high.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a block diagram of a conventional apparatus for encoding an image;

FIG. 2 is a block diagram of an apparatus for encoding an image, according to an embodiment of the present invention;

FIG. 3 is a block diagram of a residual block generator included in the apparatus for encoding an image of FIG. 2, according to an embodiment of the present invention;

FIG. 4 is a block diagram of a lossless encoding unit included in the apparatus for encoding an image of FIG. 2, according to an embodiment of the present invention;

FIGS. 5A through 5C illustrate interblock prediction, according to an embodiment of the present invention;

FIGS. 6A and 6B illustrate algorithms of interblock prediction in the apparatus for encoding an image of FIG. 2, according to an embodiment of the present invention;

FIG. 7A and 7B illustrate optimization, according to an embodiment of the present invention;

FIG. 8 illustrates an algorithm of optimization, according to an embodiment of the present invention;

FIG. 9 is a flowchart illustrating a method of encoding an image, according to an embodiment of the present invention;

FIG. 10 is a block diagram of an apparatus for decoding an image, according to an embodiment of the present invention;

FIG. 11 is a block diagram of a lossless decoding unit included in the apparatus for decoding an image of FIG. 10, according to an embodiment of the present invention;

FIGS. 12A and 12B illustrate algorithms of interblock estimation in the apparatus for decoding an image of FIG. 10, according to an embodiment of the present invention; and FIG. 13 is a flowchart illustrating a method of decoding an image, according to an embodiment of the present invention.

BEST MODE

Hereinafter, the present invention will be described more fully with reference to the accompanying drawings, in which exemplary embodiments of the invention are shown.

FIG. 2 is a block diagram of an apparatus for encoding an image, according to an embodiment of the present invention.

Referring to FIG. 2, the apparatus for encoding an image 200 includes a prediction unit 210, a subtraction unit 220, a residual block generator 230, a lossless encoding unit 240, an addition unit 250, a filter 260, and a frame memory 270. Hereinafter, encoding a predetermined block of an input image, that is, a current block will be described as an example throughout the specification of the present invention.

The prediction unit 210 predicts the current block using inter prediction or intra prediction. Also, the prediction unit 210 searches for at least one reference picture so as to perform inter prediction or generate a prediction block of the current block by performing intra prediction using pixels included in a previously encoded region that is adjacent to the current block. The subtraction unit 220 receives the prediction block from the prediction unit 210 and subtracts the estimation block from the current block, thereby generating a first residual block. The residual block generator 230 loss encodes the first residual block received from the subtraction unit 220. The residual block generator 230 is described in more detail with reference to FIG. 3.

FIG. 3 is a block diagram of the residual block generator 230 included in the apparatus for encoding an image of FIG. 2, according to an embodiment of the present invention.

Referring to FIG. 3, the residual block generator 230 includes a transformer 232, a quantization unit 234, a dequantization unit 236, and an inverse transformer 238.

The transformer 232 transforms the first residual block to the frequency domain. More preferably, the first residual block is discrete cosine transformed (DCT) to generate DCT coefficients. The quantization unit 234 quantizes the DCT coefficients generated in the transformer 232. Loss causes during quantization. The dequantization unit 236 dequantizes again the quantized DCT coefficients to restore the DCT coefficients. The inverse transformer 238 inverse discrete cosine transforms (IDCT) the restored DCT coefficients and generates the restored first residual block, that is, a second residual block. As the loss causes during quantization, the second residual block is different from the first residual block generated in the subtraction unit 220.

The lossless encoding unit 240 entropy codes the second residual block generated in the residual block generator 230 in pixel domain. According to the prior art, the quantized DCT coefficients in the frequency domain are entropy coded; however, residual values of the residual block in the pixel domain are entropy coded in the present invention.

More preferably, the lossless encoding unit 240 may transform a predetermined second residual block into a form appropriate for compression encoding through a predetermined preprocessing and then entropy codes the second residual block, instead of directly entropy coding the second residual block.

FIG. 4 is a block diagram of the lossless encoding unit 240 included in the apparatus for encoding an image of FIG. 2, according to an embodiment of the present invention.

Referring to FIG. 4, the lossless encoding unit 240 includes a preprocessor 242 and an entropy coding unit 244.

The preprocessor 242 transforms the second residual block generated in the residual block generator 230 into a form appropriate for compression encoding. This will be described in more detail with reference to FIGS. 5 and 6.

(1 ) Interblock prediction

FIGS. 5A through 5C illustrate interblock prediction, according to an embodiment of the present invention. In interblock estimation, each of residual values included in the second residual block are predicted from another residual value adjacent to a vertical or horizontal direction to the each of residual values. According to the prediction result, a third residual block is generated.

The case where the second residual block generated in the residual block generator 230 is a 4X4 block is described. FIG. 5A illustrates the second residual block, FIG. 5B illustrates the third residual block generated by performing interblock prediction in a horizontal direction, and FIG. 5C illustrates the third residual block generated by performing interblock prediction in a vertical direction.

The interblock prediction in a horizontal direction is firstly described. FIGS.5B and 6A illustrate an algorithm to perform the interblock prediction in a horizontal direction. It is assumed that the residual values of the i^th row and the j^th column of the second residual block are r(i, j) and the residual values of the i^th row and the j^th column of the third residual block are s(i, j)

Referring to FIG.6A, s(1, 1) of FIG.5B is same with r(1, 1). r(2, 1) is estimated from r(1, 1). In other words, s(2, 1) of the third residual block = r(2,

1) - r(1, 1). Similarly, r(3, 1) and r(4, 1) included in the first low are predicted for calculating s(3, 1 ) = r(3, 1 ) - r(2, 1 ) and s(4, 1 ) = r(4, 1 ) - r(3, 1 ). Then, the residual values in second, third, and fourth columns included in each row are estimated in a horizontal direction. As an example, the first row is calculated as follows. s(1,2) = r(1,2)-r(1,1) s(1,3) = r(1,3)-r(1,2) s(1,4) = r(1,4)-r(1,3)

Starting from the residual values of the first column, the residual values of other rows are estimated in a horizontal direction.

The interblock estimation in a vertical direction is described. FIGS.5C and 6B illustrate an algorithm to perform the interblock estimation in a vertical direction.

Referring to FIG.6B, s(1,1) of FIG.5C is same with r(1, 1). r(1, 2) is predicted from r(1, 1). In other words, s(1, 2) of the third residual block = r(1,

2) - r(1, 1). Similarly, r(1, 3) and r(1, 4) included in the first row are estimated for calculating s(1 , 3) = r(1 , 3) - r(1 , 2) and s(1 , 4) = r(1 , 4) - r(1 , 3). Then, the residual values in second, third, and fourth rows included in each column are estimated in vertical direction. As an example, the first row is calculated as follows. s(2, 1) = r(2, 1)-r(1, 1) s(3, 1) = r(3, 1)-r(2, 1) s(4, 1) = r(4, 1)-r(3, 1)

Starting from the residual values of the first row, the residual values of other columns are estimated in a vertical direction.

(2) Optimization The preprocessor 242 may perform not only the interblock prediction but also transform the second residual block using optimization which will be described later. The optimization is described with reference to FIGS. 7 and 8.

FIG. 7A illustrates the first residual block generated in the reduction unit 220. When the second residual block generated in the residual block generator 230 is as of FIG. 5A₁ the preprocessor 242 combines the first residual block and the second residual block to generate a fourth residual block illustrated in FIG. 7B. The residual values of the first residual block and the second residual block are compared and the fourth residual block appropriate for compression encoding is generated.

FIG. 8 illustrates an algorithm of optimization, according to an embodiment of the present invention.

Referring to FIG. 8, the fourth residual block of FIG. 7B is generated by comparing the residual values of the first residual block and the second residual block in the ratio of one to one. The residual value of the first residual value r'(1 , 1) and the residual value of the second residual value r(1 , 1 ) are compared to select the residual value having the small absolute value. All residual values in the residual blocks are compared in the ratio of one to one to select only the residual values having the small absolute values.

In the fourth residual block of FIG. 7B, the residual values of (1 , 2) (2, 1 ) (2, 3) (3, 2) (3, 4) and (4, 3) are the residual values of the first residual block and the remaining residual values are the residual values of the second residual block.

The preprocessor 242 may only perform interblock prediction described above to transform the second residual block or may only perform the optimization to transform the second residual block. In addition, the preprocessor 242 may perform both interblock estimation and the optimization to transform the second residual block. In other words, the preprocessor 242 performs the optimization first to generate the fourth residual block and performs the interblock prediction for the fourth residual block to generate the third residual block.

The entropy coding unit 244 entropy codes the third residual block or the fourth residual block generated as a result of the preprocessing performed by the preprocessor 242. When only the optimization is performed, the fourth residual block is entropy coded. When only the interblock prediction is performed and the interblock prediction is performed after the optimization, the third residual block is entropy coded. FIG. 9 is a flowchart illustrating a method of encoding an image, according to an embodiment of the present invention.

Referring to FIG. 9, in operation 910, the apparatus for encoding an image illustrated in FIG. 2 according to an embodiment of the present invention generates the first residual block of the current block using intra prediction or inter prediction. The first residual block is generated by subtracting the prediction block from the current block, wherein the prediction block of the current block of is generated using intra prediction or inter prediction.

In operation 920, the first residual block generated in operation 910 is loss encodes in the frequency domain. The first residual block is discrete cosine transformed (DCT) to generate DCT coefficients and the generated DCT coefficients are quantized.

In operation 930, the apparatus for encoding an image dequantizes again and inverse discrete cosine transforms (IDCT) the quantized DCT coefficients to generate the second residual block of the pixel domain. The second residual block includes the loss and thus is different from the first residual block, since the second residual block is decoded after loss encoding.

In operation 940, the apparatus for encoding an image entropy codes the second residual block restored in operation 930. A predetermined preprocessing may be performed for the second residual block and then the second residual block may be entropy coded. Also, the interblock prediction may be performed for the second residual block to generate the third residual block and then the third residual block may be entropy coded. Moreover, the second residual block may be optimized according to the algorithms illustrated in FIGS. 7 and 8 to generate the fourth residual block and the fourth residual block may be entropy coded. Both the optimization and interblock prediction may be performed to transform the second residual block and then the transformed second residual block may be entropy coded.

FIG. 10 is a block diagram of an apparatus for decoding an image, according to an embodiment of the present invention.

Referring to FIG. 10, the apparatus for decoding an image 1000 includes a lossless decoding unit 1010, a restoring unit 1020, a filter 1030, a frame memory 1040, and a prediction unit 1050.

The lossless decoding unit 1010 entropy decodes data for the residual block of the current block and restores the residual block. The lossless decoding unit 1010 receives the data of the residual block of the pixel domain entropy encoded by the apparatus for encoding an image illustrated in FIG. 2. And the lossless decoding unit 1010 entropy decodes the received data to restore the residual block. FIG. 11 is a block diagram of the lossless decoding unit 1010 included in the apparatus for decoding an image 1000 of FIG. 10, according to an embodiment of the present invention.

Referring to FIG. 11 , the lossless decoding unit 1010 includes an entropy decoding unit 1012 and a postprocessor 1014. The entropy decoding unit 1012 receives data for the residual block of the current block and entropy decodes the received data. As a result of the entropy decoding, the third residual block described with reference to FIGS. 5B and 5C or the fourth residual block described with reference to FIG. 7B is restored.

In case of the fourth residual block generated by performing only the optimization described with reference to FIGS. 7A and 7B, the fourth residual block, as it is, becomes the residual block of the current block so that any post processing is not needed. However, in the case of the third residual block generated by performing interblock prediction, a separate postprocessing is needed. The interblock prediction is described with reference to FIGS. 12A and 12B.

FIG. 12A illustrates an algorithm to restore the residual block using interblock prediction in a horizontal direction. When the third residual block decoded in the entropy decoding unit 1012 is the block preprocessed using interblock prediction in a horizontal direction illustrated in FIG. 5B, a postprocessing is performed for the third residual block using the same method that is symmetrical with the interblock prediction in a horizontal direction of FIG. 5B and the residual block of the current block can be restored.

Accordingly, the residual value r(i, j) of i^th row and j^th column of the residual block is calculated from the residual value s(i, j) of the third residual block as follows. Firstly, r(1 , 1) is s(1 , 1). Then, r(2, 1 ) is obtained by adding s(2, 1 ) and r(1 , 1 ). Similarly, r(3, 1 ) and r(4, 1 ) are restored. When the residual values of the first column of all rows are restored, the remaining residual values are restored using interblock prediction in a horizontal direction. As an example of the first row, the remaining residual values can be restored as follows. r(1 , 2) = s(1 , 2) + r(1 , 1) r(1 , 3) = s(1 , 3) + r(1 , 2) r(1 , 4) = s(1 , 4) + r(1 , 4) FIG. 12B illustrates an algorithm to restore the residual block using interblock prediction in a vertical direction. When the third residual block decoded in the entropy decoding unit 1012 is the block preprocessed using interblock prediction in a vertical direction illustrated in FIG. 5C, a postprocessing is performed for the third residual block using the same method that is symmetrical with the interblock prediction in a vertical direction of FIG. 5C and the residual block of the current block can be restored.

Accordingly, the residual value r(i, j) of i^th row and j^th column of the residual block is calculated from the residual value s(i, j) of the third residual block as follows. Similarly, r(1 , 3) and r(1 , 4) are restored. When the residual values of the first row of all columns are restored, the remaining residual values are restored using interblock prediction in a vertical direction. As an example of the first column, the remaining residual values can be restored as follows. r(2, 1 ) = s(2, 1 ) + r(1 , 1) r(3, 1 ) = s(3, 1 ) + r(2, 1 ) r(4, 1 ) = s(4, 1 ) + r(3, 1 )

Referring back to FIG. 10, the restoring unit 1020 restores the current block based on the residual block of the current block generated by entropy coding and postprocessing in the lossless encoding unit 1010. The restoring unit 1020 adds the prediction block of the current block generated by performing intra prediction or inter prediction in the prediction unit 1050 to the residual block to store the current block.

The restored current block is deblocking filtered in the filter 1030 and the deblocking filtered current block is stored in the frame memory 1040 for estimation of the next picture or the next block.

FIG. 13 is a flowchart illustrating a method of decoding an image, according to an embodiment of the present invention.

Referring to FIG. 13, in operation 1310, the apparatus for decoding an image illustrated in FIG. 10 according to an embodiment of the present invention entropy decodes data for the residual block of the current block and restores the residual block. The apparatus for decoding an image receives the data of the residual block of the pixel domain entropy encoded by the method of encoding an image according to the present invention and decodes the received data to restore the residual block. In case of the third residual block generated by performing interblock prediction, a separate postprocessing is needed. That is, the postprocessing is performed using the interblock prediction in a horizontal direction or in a vertical direction described with reference to FIGS. 12A and 12B. As a result of the postprocessing, the residual block of the current block is restored from the third residual block.

In operation 1320, the apparatus for decoding an image illustrated in FIG. 10 restores the current block based on the residual block restored in operation 1310. The prediction block of the current block is generated by performing intra prediction or inter prediction and the generated prediction block is added to the residual block restored in operation 1310 to store the current block.

The invention can also be embodied as computer readable codes on a computer readable recording medium. The computer readable recording medium is any data storage device that can store data which can be thereafter read by a computer system. Examples of the computer readable recording medium include read-only-memory (ROM), random-access memory (RAM), CD-ROMs, magnetic tapes, floppy disks, optical data storage devices, and carrier waves (such as data transmission through the Internet). The computer readable recording medium can also be distributed over network coupled computer systems so that the computer readable code is stored and executed in a distributed fashion.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it will be understood by those of ordinary skill in the art that various changes in form and details may be made therein without departing from the spirit and scope of the present invention as defined by the following claims.

Claims

1. A method of encoding an image comprising: generating a first residual block of a current block using intra prediction or inter prediction; loss encoding the first residual block in frequency domain; decoding the encoded first residual block and restoring a second residual block in pixel domain; and entropy coding the restored second residual block.

2. The method of claim 1 , wherein the loss encoding comprises: discrete cosine transforming (DCT) the first residual block and generating

DCT coefficients; and quantizing the generated DCT coefficients.

3. The method of claim 2, wherein the entropy coding comprises: predicting each of residual values included in the restored second residual block from another residual value adjacent to a vertical direction to the each of residual values or another residual value adjacent to a horizontal direction to the each of residual values and generating a third residual block based on the result of the prediction; and entropy coding the generated third residual block.

4. The method of claim 2, wherein the entropy coding comprises: comparing the residual values of the first residual block and the second residual block and generating a fourth residual block only including the residual values having the small absolute value; predicting each of residual values of the fourth residual block from another residual value adjacent to a vertical direction to the each of residual values of the fourth residual block or another residual value adjacent to a horizontal direction to the each of residual values and generating a third residual block based on the result of the prediction of the residual values included in the fourth residual block; and entropy coding the generated third residual block.

5. A method of decoding an image comprising: entropy decoding data for a residual block of a current block and restoring the residual block; and restoring the current block based on the restored residual block, wherein the data for the residual block is for a second block restored by decoding a first residual block loss encoded in frequency domain, and wherein the first residual block of the current block is generated by performing intra prediction or inter prediction while encoding the current block.

6. The method of claim 5, wherein the restoring comprises: entropy decoding data for the residual block of the current block and restoring residual values of a third residual block; restoring residual values of the second residual block using prediction residual values predicted from another residual value adjacent to a vertical direction to each of the prediction residual block or another residual value adjacent to a horizontal direction to each of the residual values of prediction residual block and the residual values of the third residual block.

7. An apparatus for encoding an image comprising: a prediction unit which generates an prediction block of a current block using intra prediction or inter prediction; a subtracting unit which generates a first residual block by subtracting the generated prediction block from the current block; a residual block generator which loss encodes the first residual block in frequency domain, decodes the encoded first residual block, and restores a second residual block of pixel domain; and a lossless encoding unit which entropy encodes the restored second residual block.

8. An apparatus for decoding an image comprising: a lossless decoding unit which entropy decodes data for a residual block of a current block and restores the residual block; and a restoring unit which restores the current block based on the restored residual block, wherein the data for the residual block is for a second block restored by decoding a first residual block loss encoded in frequency domain, and wherein the first residual block of the current block is generated by performing intra prediction or inter prediction while encoding the current block.

9. A computer readable recording medium having embodied thereon a computer program for executing the method of any one of claims 1-6.