KR100670459B1 - Image compression coding and decoding method and apparatus - Google Patents

Abstract

The present invention relates to an apparatus and method for compressing and encoding an image.

According to the present invention, in processing a digital image signal based on a macro block, a seed pixel is selected for each macro block, a difference image between the seed pixel and its surrounding pixels is formed, and the input image is subsampled at a specific ratio. A sub-image is formed to form a seed image, the seed image and the difference image are DCT and quantized, respectively, and compressed, and the image is compressed by lossless coding. The encoded data of the seed image and the difference image are respectively decompressed in the reverse order of the encoding process, and in the final step, two images are added to obtain a decoded image. According to the present invention, when compression is performed using only information in a frame, compression efficiency can be improved by removing spatial redundancy in an image.

Video compression, video encoding, MPEG, CODEC

Description

Video compression encoding and decoding method and apparatus {VIDEO CODING AND DECODING SYSTEM}

1 is a block diagram of a conventional digital video encoder.

2 is a flowchart showing a video compression encoding and decoding method according to the present invention.

3 is a view showing an example of the configuration of the seed pixel and the seed image in the present invention

4 shows an example of the seed line configuration in the present invention.

5 is a view showing an example of the pixel difference value configuration between lines in the present invention;

6 is a block diagram showing an image compression encoding and decoding apparatus according to the present invention.

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

610: image encoder 620: image decoder

611: difference image generation unit 612,616: DCT unit

613,617 quantization unit 614,618 lossless coding unit

615: Seed image generation unit 619: Adder

621,625: Lossless decoding unit 622,626: Inverse quantization unit

623,627: reverse DCT 624: restored car image

628: Restored seed image 629: Adder

The present invention relates to an apparatus and method for compressing and encoding an image.

Most video compression standards, such as MPEG or H.26x video codecs, adopt compression based on motion estimation and compensation and transformation. In such motion compensation based encoding, motion vector information of each block must be encoded and transmitted, and the compression efficiency may vary greatly depending on how the motion vector is encoded.

In the general process of encoding an image, DCS (Discrete Cosine Transform) processing of a digital image signal is performed, and the transform coefficient is quantized to perform lossless coding, while inverse quantization and inverse DCT transformation of the quantized DCT coefficients are performed. The image is stored in a memory, a motion vector is calculated using the reconstructed image and the next frame image stored in the memory, and the motion vector is variably encoded to form a bit stream along with the encoded image information for transmission. The method of decoding an image is performed in the reverse order of the encoding process.

1 is a block diagram showing the configuration of a general digital video encoder.

The video coder illustrated in FIG. 1 includes a DCT unit 110 for processing a DCT (Discrete Cosine Transform), a quantizer 120 for quantizing the DCT transform coefficients, and an output from the quantizer 120. An inverse quantization unit 130 for inverse quantization of quantization data, an inverse DCT unit 140 for inverse DCT transforming the output data of the inverse quantization unit 130, and a reconstructed image output from the inverse DCT unit 140 A motion estimator 160 that calculates a motion vector (MV) by using the frame memory 150 to be stored, the image data of the current frame to be input, and the image data of the previous frame stored in the frame memory 150, and And a lossless encoder 170 for lossless coding and outputting the quantized data and the motion vector.

Looking at the encoding process of the compression encoder of Figure 1 configured as described above is as follows.

The DCT (Discrete Cosine Transform) unit 110 performs DCT transformation on the digital image signal input in units of 8 * 8 pixel blocks, and the quantization unit (Q) 120 is obtained by the DCT unit 110. High efficiency compression is performed by performing quantization on the DCT coefficients and expressing them as some representative values.

An inverse quantization unit (IQ) 130 inverse quantizes the quantized data output from the quantization unit 120, and inversely quantizes the inverse quantized data by the inverse quantization unit 130 in the IDCT unit 140. Perform a discrete cosine transform (Inverse DCT). The frame memory 150 stores IDCT-converted image data by the IDCT unit 140 in units of frames.

The motion estimation unit (ME) 160 calculates a motion vector per macroblock (MV) by using the input image data of the current frame and the image data of the previous frame stored in the frame memory 150. The lossless encoding unit 170 receives the quantized output data and the motion vector MV from the quantization unit 120, and then outputs them by lossless coding.

As is known, MPEG is a representative algorithm for compressing a video using time redundancy. In the B frame or P frame of MPEG, the temporal redundancy with the previous frame can be efficiently removed by using the motion vector of the macro block. However, when compressing using only the information inside the frame, such as MPEG I frame or JPEG, time redundancy cannot be used. In this case, most of the compression efficiency can be increased by decomposing the input digital video signal by frequency using DCT and removing data in the high frequency region which is visually insensitive. However, the compression of information using DCT has a limitation that the compression efficiency is inevitably reduced because it does not remove the spatial redundancy that occupies a large amount of data in the actual image. In the case of MPEG described above, it can be understood from the fact that the information amount of P frame and B frame is significantly smaller than that of I frame using only internal information.

Therefore, when compressing using only information in a frame, a technique for further increasing compression efficiency by removing spatial redundancy in an image is required.

SUMMARY OF THE INVENTION An object of the present invention is to provide an image compression encoding and decoding method and apparatus for solving the problems of the related art described above, thereby improving encoding efficiency of an image compression encoder.

Another object of the present invention is to provide an image compression encoding and decoding method and apparatus for improving compression encoding efficiency by compressing by using spatial information redundancy of an image during image compression encoding.

It is still another object of the present invention to provide an image compression encoding and decoding method and apparatus for improving compression efficiency by removing spatial redundancy in an image when compression is performed using only information in a frame during image compression encoding. It is.

It is still another object of the present invention to select a pixel that becomes a seed in a macro block based on image compression coding in units of blocks, and to neighbor the neighbors around the selected seed pixel. Obtains a differential image obtained by calculating a difference between pixels and a seed image sub-sampled at a specific ratio, and transforms and encodes the seed image and the differential image, respectively, The present invention provides an image compression encoding method and apparatus for removing spatial redundancy as much as possible and thereby achieving high compression efficiency.

Another object of the present invention is to decode and inverse transform the compressed coded seed image and the difference image provided in the encoding step, respectively, based on image compression decoding in units of blocks, and add the two images decoded in the final step. The present invention provides a compressed image decoding method and apparatus for decoding an image by a method of restoring an original image.

In the video compression encoding method of the present invention for achieving the above object, the image data is divided into block units, a pixel of a predetermined position is selected within each block, and the seed data is selected as a seed pixel. Seed pixel selection step; The difference between neighboring neighboring pixels is determined based on the selected seed pixel to form a differential image including the difference, and sub-sampling the image data at a predetermined ratio. Constructing a difference image and a seed image to determine a seeded image as a seed image; An encoding step of performing compression encoding on each of the difference image and the seed image data, and then adding the two compression-encoded data and outputting the final compression-encoded data; Characterized in that comprises a.

In addition, the video decoding method of the present invention for achieving the above object, the seed image obtained by sub-sampling the image data, and the difference between the neighboring pixels based on the seed pixel selected for each block by dividing the image data in block units Receiving image data obtained by compressing and encoding the difference image data; Decoding the compressed seed image data and the differentially encoded difference image data, decompressing each other to obtain original seed image data and difference image data, and adding the obtained seed image data and difference image data to obtain a decoded image. ; Characterized in that it comprises a.

In addition, according to an embodiment of the present invention, an apparatus for encoding and compressing image data based on a block unit is provided. The apparatus for encoding and compressing image data based on a block unit may include: Difference image generation means for constructing a difference image comprising a difference; Seed image generating means for sub-sampling the image data at a predetermined ratio to form a seed image; Differential image compression encoding means for sequentially performing DCT, quantization, and lossless encoding on the difference image; Seed image compression encoding means for performing DCT, quantization, and lossless encoding on the seed image in sequence; Output means for adding and outputting the finally encoded difference image and the seed image; Characterized in that configured to include.

Also, an image decoding apparatus of the present invention for achieving the above object is a device for decoding compressed image data based on a block unit, comprising: between a seed pixel selected in an image data block and neighboring pixels adjacent to the periphery thereof; Difference image reconstruction means for decoding and reconstructing a seed pixel-based difference image by sequentially performing decoding, inverse quantization, and inverse DCT on a compression-coded difference image including a difference; Seed image reconstruction means for decoding and reconstructing the seed image by sequentially performing decoding, inverse quantization, and inverse DCT processes on the compressed coded seed image configured by subsampling image data at a predetermined ratio; Output means for reconstructing the original image by adding the finally reconstructed difference image and seed image; Characterized in that configured to include.

In another aspect, the present invention is characterized by dividing the image data into macroblock units, selecting one pixel located at the center of each macroblock, and selecting the pixel as a seed pixel.

In addition, the present invention forms a seed line including a seed pixel by obtaining the difference between neighboring pixels adjacent to the left and right adjacent to the seed pixel, and the same neighboring up and down adjacent to the seed line The difference image may be obtained by obtaining the difference between the pixels of the column position.

In addition, the present invention forms a seed column including a seed pixel by obtaining a difference between neighboring pixels vertically adjacent to the seed pixel up and down, and the same neighboring left and right adjacent to the seed column The difference image may be obtained by obtaining a difference between pixels at a line position.

In the present invention, the difference image is characterized in that the hierarchical structure.

The present invention also divides the seed image into predetermined block units, selects seed pixels in each block, and repeats a process of obtaining a difference between neighboring pixels adjacent to the seed pixels. It is characterized by forming a structure.

In addition, in the present invention, the seed image may be configured by subsampling to include the selected seed pixel.

According to the image compression encoding and decoding method and apparatus thereof according to the present invention made as described above, a difference image obtained by calculating a difference between neighboring pixels centering on a pixel that is a seed in a macroblock. (differential image) is obtained. This difference image can improve compression efficiency because the redundancy in two-dimensional space is removed.

In addition, according to the video compression encoding and decoding method and apparatus according to the present invention as described above, it is possible to further increase the efficiency of compression by enabling to produce a difference image of a variable step. This means that the bit-rate of data can be adjusted.

In addition, according to the video compression encoding and decoding method and apparatus of the present invention as described above, it is possible to reduce the deterioration of the image caused by the compression by making it possible to produce a difference image of various stages.

Further, according to the video compression encoding and decoding method and apparatus of the present invention made as described above, the present invention can be applied to video compression. That is, the compression efficiency can be increased by applying the same to an I frame that is compressed using only the information inside the frame in MPEG or the like.

Hereinafter, with reference to the accompanying drawings will be described an embodiment of the present invention.

First, FIG. 2 shows a video compression encoding and decoding method according to the present invention.

In operation S211, the seed pixels are selected and a seed image is constructed. The seed pixel divides the digital image data into block units and selects each block within the block. For example, when processing by dividing into macroblock units, a pixel is selected and selected in each macroblock for each macroblock. That is, a pixel located in the center of the macro block may be defined as a seed pixel. Which pixel to select as the seed pixel for the macroblock may be not limited. The selection of the seed pixel may vary depending on the characteristics, communication, and application environment of the system to which the present invention is applied. In general, however, it is preferable to select one pixel located at the center of the macro block as a seed pixel. The reason for this is that if the position of the seed pixel is promised in advance, not only is it convenient for encoding and decoding, but also the pixel located in the center of the macro block can be considered to best represent the information of the macro block. . 3 shows a source image 300 and a macro block 310, a seed pixel 311 selected at the center of the macro block 310, and a seed image 320 obtained through subsampling. Shows.

The subsampled image is obtained by subsampling the image data at a predetermined sampling rate. The subsampled image is defined as the seed image 320. The seed image 320 may or may not include the seed pixel. This depends on the subsampling technique. However, in general, it is advantageous to configure the seed image 320 by including the seed pixel 311 to include richer information about the image data.

In FIG. 2, the next step S212 is to construct a difference image. Here, the difference image is configured by obtaining a difference from the neighboring pixels based on the seed pixel 311 and processed in units of macro blocks. 4 and 5 show an example of obtaining a difference image. That is, a seed line including a seed pixel is formed by repeating a process of obtaining difference values between neighboring pixels on the left and right sides based on the seed pixel, and neighboring the top and bottom sides based on the seed line. The difference image is generated by repeating the process of obtaining the difference value between the pixels.

First, reference is made to FIG. 4 shows a method of obtaining a difference value between neighboring pixels adjacent to the seed pixel P (0,0) from left to right. Difference between pixel P (-2,0) and pixel P (-1,0) based on seed pixel P (0,0), pixel P (-1,0) and seed pixel P (0,0) ), The difference between the seed pixel P (0,0) and the pixel P (1,0), and the difference between the pixel P (1,0) and the pixel P (2,0). For example, if the image input corresponding to each pixel is [128,120,134,140,120], the difference image is [8, -14,0,6, -20]. After this process, the line including the seed pixel P (0,0) becomes a seed line.

Next, reference is made to FIG. 5. 5 shows a method of obtaining a difference value between neighboring pixels adjacent to the seed line in both up and down directions. When the difference between lines is obtained, the difference between pixels at the same column position of both lines is obtained. For example, in the seed line, the difference between the pixel P (-2,0) and the neighboring pixel P (-2,1) upwards is obtained, and the pixel P (-1,0) and the pixel P (-1,1) Find the difference value of, find the difference between the seed pixel P (0,0) and the pixel P (0,1), find the difference between the pixel P (1,0) and the pixel P (1,1), and The difference between P (2,0) and pixel P (2,1) is obtained. In the same way, the difference between pixel P (-2,1) and pixel P (-2,2) is calculated, the difference between pixel P (-1,1) and pixel P (-1,2) is obtained, and the pixel Find the difference between P (0,1) and pixel P (0,2), find the difference between pixel P (1,1) and pixel P (1,2), and find pixel P (2,1) and pixel Find the difference between P (2,2). In this way, the difference between pixels neighboring the same column of all the remaining lines is obtained. In this way, the difference image is constructed.

In this way, one seed image and multiple layers of difference images may be included in one frame. That is, if the sub-sampled seed image is regarded as an input image again and the above series of processes are repeated, a difference image having a hierarchical structure can be produced.
In other words, the seed image is defined as a source image, divided into predetermined block units, one seed pixel is selected from each of the divided blocks, and the above-described drawing is performed based on the selected seed pixel. In the same manner as shown in Figs. 4 and 5, difference images between adjacent neighboring pixels are obtained.
Repeating this process n times may have an n-layer difference image structure. That is, it is possible to create a difference image of various layers having a correlation with the seed image subsampled in the frame.
By making the difference image of variable stages as described above, the efficiency of compression can be further increased, which means that the bit rate of the image data can be adjusted. That is, the bit rate is selected (adjusted) according to which layer difference image is compressed and output from the hierarchical difference image. In addition, since it is possible to produce a difference image of various stages, it is possible to reduce the loss of image information as much as possible to reduce the deterioration of the image caused by the compression.

4 and 5 illustrate a method of constructing a difference image by obtaining a difference between neighboring pixels of a seed line including a seed pixel and a difference between pixels of the same column position of adjacent seed lines. However, on the contrary, the neighboring pixels P (0,2), P (0,1), P (0,0), P (0, -1) It is also possible to find the difference between the adjacent columns on both the left and right sides of the seed column by finding the difference between P (0, -2) and using this as the seed column. Of course, when determining the difference between columns, the difference between pixels at the same line position of both lines is obtained.

As described above, whether to construct a seed line based on the seed pixel and obtain adjacent pixel differences between the lines, or to configure a seed column based on the seed pixel and obtain adjacent pixel differences between the columns What is necessary is just to make it suitable according to the environment of a device, communication, a medium, etc. which apply an image | video encoding and a decoder.

Referring back to FIG. 2, a next step S213 is compressing and encoding the seed image and the difference image. That is, as described above, the seed image obtained by subsampling the original image data is discrete cosine transformed (DCT), and the DCT coefficients are quantized, and then the lossy coding (Variable Length Coding (VLC)) is performed again. compression coding on the seed image by lossless coding, discrete cosine transform (DCT) of the difference image data, quantize the DCT coefficients, and then perform lossless coding such as variable length encoding to perform compression encoding on the difference image. Perform. In addition, the compressed and encoded seed image and the difference image may be finally output in correspondence with each other, thereby outputting compressed image data having the maximum spatial redundancy.

That is, according to the image compression encoding method of the present invention, since the image entering the input of the DCT is a difference image, the redundancy in the space is removed, thereby increasing the compression efficiency.

The method of decoding the compression-encoded image data includes a seed image and difference image decoding step S311, a seed image and difference image compression restoring step S312, and a configuration and output step of a decoded (decoded) image in FIG. 2. )

First, in the seed image and difference image decoding step (S311), the compression-coded seed image is losslessly decoded (inverse lossless coding = lossless decoding), and the compression-coded difference image is also losslessly decoded. In the next step (S312), the decompression is performed by inverse quantization and inverse DCT on each of the decoded difference image and the seed image. In this way, by adding the reconstructed difference image and the seed image in the final step (S313), it is possible to construct and obtain the reconstructed image.
For example, if the seed image is composed of seed pixels located at the center of each macro block of the original image data, the seed image is already obtained as shown in FIG. 4 and FIG. It is possible to reconstruct the pixel value of the original image data by adding the difference image, which is a difference value between neighboring pixels, in the same manner.

6 shows a configuration of an embodiment of an image compression encoding and decoding apparatus according to the present invention. Referring to FIG. 6, an image compression encoding and decoding apparatus (CODEC) according to the present invention includes an image encoder 610 and an image decoder 620.

The image encoder 610 performs discrete cosine transform on a difference image generator 611 for obtaining a difference image of an input source image, and a difference image generated by the difference image generator 611. A first DCT unit 612, a first quantizer 613 for quantizing the DCT coefficients output from the first DCT unit 612, and a first lossless coding quantization data of the first quantizer 613 1 A lossless encoder 614, a seed image generator 615 obtaining a seed image of the original image, and a second cosine transform on the seed image generated by the seed image generator 615. A second lossless encoding lossless encoding of the DCT unit 616, the second quantization unit 617 for quantizing the DCT coefficients output from the second DCT unit 616, and the quantization data of the second quantization unit 617. The encoder 618 outputs the first lossless encoder 614 and the second lossless encoder 618. And an adder 619 which adds and finally outputs the compressed coded image data.

The image decoder 620 may include a first lossless decoder 621 which performs inverse lossless coding on the difference image data among the compressed coded data, and data output from the first lossless decoder 621. A first inverse quantizer 622 for inverse quantization and a first inverse DCT for inverse discrete cosine transforming data output from the first inverse quantizer 622 to output a difference image 624. An inverse DCT) unit 623, a second lossless decoding unit 625 that losslessly decodes seed image data among the compressed coded data, and an inverse quantization of data output from the second lossless decoding unit 625 A second inverse quantization unit 626, a second inverse DCT unit 627 that outputs a seed image 628 by inverse discrete cosine transforming the data output from the second inverse quantization unit 626, and the difference Decoded by adding image 624 and seed image 624 Phase includes the addition unit 629 to finally output an (decode image).

The video compression encoding and decoding process by the video encoding and decoding apparatus of the present invention configured as described above is performed in the same manner as the above-described video compression encoding and decoding method.

The difference image generation unit 611 receives the source image data and configures the difference image as described with reference to FIGS. 4 and 5 on the basis of the seed pixel selected for each macro block. It has already been explained that this difference image can be generated in a hierarchical structure. In addition, since the spatial redundancy is removed in this difference image, the compression efficiency thereof becomes much higher than that of the conventional compression method during the subsequent compression encoding process. The difference image data generated by the difference image generation unit 611 is input to the first DCT unit 612 and is discrete cosine transformed. The DCT coefficients output from the first DCT unit 612 are quantized by the first quantizer 613, and the quantized data is losslessly coded by the first lossless encoder 614 and supplied to the adder 619. The lossless coding may include, for example, run length coding, variable length coding, and the like.

The seed image generator 615 receives the source image data and subsamples the image data at a specific ratio to configure the seed image as illustrated in FIG. 3. The seed image data generated by the seed seed image generator 615 is input to the second DCT unit 616 to be discrete cosine transformed. The DCT coefficients output from the first DCT unit 616 are quantized by the second quantizer 617, and the quantized data is losslessly encoded by the second lossless encoder 618 and supplied to the adder 619. The lossless coding may include, for example, run length coding, variable length coding, and the like.

The adder 619 adds the compression-encoded difference image and the seed image data and finally outputs the compression-encoded data.

The first lossless decoding unit 621 performs lossless decoding of the difference image data among the compressed coded data. The first inverse quantizer 622 inverse quantizes the output data of the first lossless decoder 621, and the first inverse DCT 623 outputs the first inverse quantizer 622. Inverse DCT is performed on the extracted data to restore the original difference image and output the reconstructed difference image 624. The second lossless decoder 625 losslessly decodes the seed image data from the compressed coded data. The second inverse quantization unit 626 inverse quantizes the data output from the second lossless decoding unit 625, and the second inverse DCT unit 627 in the second inverse quantization unit 626. The inverted cosine transform (inverse DCT) of the output data restores the original seed image and outputs the restored seed image 628. The adder 629 adds the reconstructed difference image 624 and the seed image 628 to output a finally decoded (decoded) image.

In particular, the image codec according to the present invention can increase compression efficiency by maximally eliminating spatial redundancy within an image. In addition, it is possible to make the difference image of hierarchical structure, that is, the hierarchical structure, so that the compression efficiency can be improved and the bit rate can be adjusted freely according to the system environment. There is an effect that can be suppressed. In particular, when applied to video compression, the compression efficiency can be increased in the case of an I frame that performs compression using only information inside a frame in MPEG or the like.

Claims (13)

A seed pixel selecting step of dividing the image data into block units, selecting a pixel at a predetermined position in each block, and selecting the pixel as a seed pixel; The difference between neighboring neighboring pixels is determined based on the selected seed pixel to form a differential image including the difference, and sub-sampling of the original image data at a predetermined ratio is performed. Constructing a difference image and a seed image to determine a sampled image as a seed image; An encoding step of performing compression encoding on each of the difference image and the seed image data, and then outputting the final compression-encoded data by matching the two compression-encoded data; Image compression encoding method comprising a. The method of claim 1, wherein in the seed pixel selection step, image data is divided into macroblock units, and one pixel located at the center of each macroblock is selected to select the pixel as a seed pixel. . The method of claim 1, wherein in the difference image and seed image forming step, a seed line including a seed pixel is formed by obtaining a difference between neighboring pixels adjacent to the left and right around the seed pixel. The image compression encoding method of claim 1, wherein the difference image is formed by obtaining a difference between pixels at adjacent column positions up and down based on the seed line. The method of claim 1, wherein in the difference image and seed image forming step, a seed column including a seed pixel is configured by obtaining a difference between neighboring pixels vertically adjacent to the seed pixel. And a difference image is formed by obtaining a difference between pixels at adjacent same line positions adjacent to the left and right based on the seed column. The method of claim 1, wherein the difference image has a hierarchical structure. 2. The method of claim 1, wherein the difference image is divided by repeating a process of dividing the seed image into predetermined block units, selecting seed pixels in each block, and obtaining a difference between adjacent pixels adjacent to the seed pixel. A video compression coding method comprising a hierarchical structure. The image compression encoding method of claim 1, wherein the seed image is configured by subsampling to include the selected seed pixel. Receiving image data obtained by sub-sampling the image data and compressing and encoding the image data by dividing the image data into units of blocks and difference image data consisting of differences between neighboring pixels based on the selected seed pixels for each block. ; Decoding the compressed seed image data and the differentially encoded difference image data, decompressing each other to obtain original seed image data and difference image data, and adding the obtained seed image data and difference image data to obtain a decoded image. ; Image decoding method comprising a. An apparatus for compressing and encoding image data based on a block unit, Difference image generation means for constructing a difference image comprising a difference between the selected seed pixel in the image data block and neighboring pixels adjacent to the periphery thereof; Seed image generating means for sub-sampling the image data at a predetermined ratio to construct a seed image; Differential image compression encoding means for sequentially performing DCT, quantization, and lossless encoding on the difference image; Seed image compression encoding means for performing DCT, quantization, and lossless encoding on the seed image in sequence; Output means for adding and outputting the finally encoded difference image and the seed image; The video compression encoding apparatus, comprising a. 10. The method of claim 9, wherein the difference image generating means forms a seed line including a seed pixel by obtaining a difference between neighboring pixels adjacent to the left and right around the seed pixel, and based on the seed line. And a difference image is formed by obtaining a difference between pixels at adjacent same column positions up and down. The method of claim 9, wherein the difference image generating unit forms a seed column including a seed pixel by obtaining a difference between neighboring pixels vertically adjacent to the seed pixel up and down, and based on the seed column. And forming a difference image by obtaining a difference between pixels at adjacent same line positions left and right adjacent to each other. 10. The method of claim 9, wherein the difference image generating means takes as input a seed image generated by the seed image generating means, divides it into predetermined block units, selects a seed pixel in each block, and adjacents the seed pixel. And repeating a process of obtaining a difference between neighboring pixels, thereby generating a difference image having a hierarchical structure. An apparatus for decoding compressed video data based on a block unit, Decoding and reconstructing the seed pixel-based difference image by performing decoding, inverse quantization, and inverse DCT on a compressed coded difference image composed of a difference between the selected seed pixel and neighboring neighboring pixels in the image data block. Difference image restoring means; Seed image reconstruction means for decoding and reconstructing the seed image by sequentially performing decoding, inverse quantization, and inverse DCT processes on the compressed coded seed image configured by subsampling image data at a predetermined ratio; Output means for reconstructing the original image by adding the finally reconstructed difference image and seed image; Image decoding apparatus characterized in that it comprises a.

Images