KR950008640B1 - Image compression coding method and decoding method for bit fixation - Google Patents

Abstract

The method codes and decodes JPEG images optimally. The method for coding JPEG image comprises the steps of: (a) coding DC coefficients in quantized block DCT coefficients according to block orders, and (b) coding AC coefficient in quantized block DCT coefficients according to position data of each of pixels and AC pixel values having large absolute values. The method for decoding coded JPEG image comprises the steps of: (a) decoding DC coefficients of each of blocks according to coded orders; (b)decoding AC coefficients according to the position data stored in position flags; and (c) decoding AC coefficients having no position data as zero.

Description

Image Compression Coding and Decoding for Bit Fixing

1 is a block diagram according to an embodiment of a video compression code and apparatus for bit fixing according to the present invention.

2 is a block diagram of an embodiment of an image compression encoding apparatus for bit fixing according to the present invention.

* Explanation of symbols for main parts of the drawings

101: color signal separation unit 102: sub-sampling unit

103: block division unit 104: DCT conversion unit

105: quantization 106, 206: quantization matrix

107, 207: Multiplier 108: DC encoding unit

109: AC encoding section 110: multiplexer

111, 201: ECC section 202: demultiplexer

203: DC decoding unit 204: AC decoding unit

205 dequantization 208 IDCT conversion unit

208: block combination unit 210: interpolation operation unit

211: color table conversion unit

The present invention relates to an image compression encoding method and a decoding method for bit fixing in a digital image device.

Generally, a variety of compression encoding methods are used to increase recording efficiency when recording color images on tape or other digital recording media.

The compression standardization method recommended by the Joint Photogrphy Expert Group (JPEG) and Moving Picture Expert Group (MPEG) by the International Organization for Standardization uses DCT variable coding. In such a variable coding scheme, it is important to fix the output bit amount according to the complexity of an image, and in particular, in the case of a digital video cassette recorder (DVCR), it is a decisive factor in trick play such as search.

An object of the present invention is to block the DC coefficient of a block that has been subjected to DCT processing for a certain coding unit in order to efficiently fix the amount of bits output efficiently. The present invention provides an encoding method and a decoding method for fixing bits transmitted by coding position information by a position plug and data values of a selected pixel for every AC coefficient of a block.

In order to achieve the above object, the video compression encoding method and the decoding method for bit fixing according to the present invention provide a block order in a DCT coefficient quantized in a plurality of block units in a DCT encoding method for compressing a digital video signal. After encoding according to the above, the AC coefficient of each block selects a pixel having a predetermined absolute value, and encodes the position information by the position flag and the data value of the selected AC coefficient for the entire AC coefficient of each block. And reconstructed to reconstruct a signal encoded by the same encoding scheme.

Hereinafter, exemplary embodiments of a video compression encoding method and a decoding method for bit fixing according to the present invention will be described with reference to the accompanying drawings.

1 is a block diagram according to an embodiment of an image compression encoding apparatus for bit fixing according to the present invention.

According to FIG. 1, the operation of the color signal separation unit 101, the quantization matrix 106, the quantization 105, and the like are described in Korean Patent Application No. 91-12412 filed by Samsung Electronics Co., Ltd. Therefore, detailed description thereof will be omitted.

The color signal separator 101 separates the R, G, and B signals into Y, which is a luminance signal, and R-Y, B-Y, which are color difference signals. The function of the subsampling unit 102 subsamples 2: 1 in the horizontal and vertical directions of the color difference signals R-Y and B-Y separated by the color signal separation unit 101. When the compressed color difference signal is restored in this way, the values between the pixels are linearly interpolated to fill the original values.

The block division unit 103 inputs the luminance signal Y output from the color signal separation unit 101 and the color difference signals RY and BY output through the subsampling unit 102 to block 8 × 8 pixels. Divide. The encoding order is processed by four blocks in the Y signal and one block in each of the R-Y and B-Y signals as one unit.

In the DCT conversion unit 104, the block of the luminance signal Y and the color difference signal RY and BY divided by the block division unit 103 is divided into blocks, that is, the rightmost left block and the lowest right side. Two-dimensional DCT transformation is performed sequentially up to the block. DCT is a type of transform encoding that compresses data by applying a quantization method suitable for the statistical properties by performing two-dimensional transform and transforming data. KLT (Karhunen- Since it has almost the same performance as Loeve Transform, it is widely used for data compression, filtering, and feature extraction.

Quantizer 105 is quantized by the output of the quantization matrix 106, quantization matrix and the quantization compression means compressing an integer (S) and a multiplier (107) multiplying the attenuation factor ^(2-5), hence from the (not shown) of the The step size is controlled. The above-mentioned both and the compression setting means are described in "91-23727" filed domestically by Samsung Electronics Co., Ltd., detailed description thereof will be omitted. In addition, the quantization matrix 106 uses a quantization matrix proposed by ISO / CCITT, an international standardization organization.

That is, Q (u, v) = QM (u, v) × S

Where QM (u, v): quantization matrix

S: quantization factor

The quantization coefficient value is quantized by the following expression.

Here, the properties of the DCT transform are classified into DC and AC components, and the farther the DC component is, the higher the frequency coefficient is. When DCT conversion of a general image, the coefficient of the low frequency component is an important factor in determining the image quality, and the magnitude of the value is much larger than the high frequency. In consideration of such factors, the quantization matrix is designed so that the low frequency components have a smaller spacing than the high frequency components. Therefore, the low frequency component near the DC component has a very high probability of being non-zero, and the probability decreases significantly toward the high frequencies. The DC encoding unit 108 expresses the DC coefficient output after the DCT conversion and quantization as an 8-bit signal.

After the DCT conversion and quantization, the AC encoding unit 109 adds position information for each of the 63 AC components except for one DC component by position flag for each pixel.

In addition, m pixels are transmitted with respect to pixels having a large absolute value, that is, important pixels, and data values for the m pixels are transmitted. In this case, m is the pixel depends on the compression rate.

Therefore, m AC values having the largest absolute value among the 63 AC coefficients are encoded for the 8x8 block. The encoded m AC values are represented by n bits, and the position flag representing position is represented by 63 bits. When the position flag is set to "1", it means that the selected absolute value is a large pixel, and is set to "0". The data value for that pixel is zero.

That is, the encoding format of the AC coefficient represents each block by 63 bits (position) + m (pixels) x n bits (value). Thus, the bit amount output per block is 8 bits (DC) + 63 bits (position) + (m (pixels) x n bits). By doing so, 64 bytes per DCT block are compressed to a fixed bit and fixed.

The multiplexer 110 alternately outputs encoding results of the DC encoding unit 108 and the AC encoding unit 109 to the Error Correction Coding (ECC) unit 111.

In the present invention, the code encoded by the coded scheme may be decoded in the reverse order as shown in FIG.

That is, the demultiplexer 201 outputs the coded DC value and the AC value output from the multiplexer 111 to the DC decoder 203 and the AC decoder 204 through the ECC unit 202. The AC decoding unit 204 decodes m pixels selected at the time of encoding among 8 × 8 blocks by position information by the position flag. That is, the position data is decoded into the transmitted data values for the m AC components whose position flag is set to "1" among 63 AC coefficients, and all the remaining pixels whose position flag is set to "0" are set to "0".

Inverse quantization 205 dequantizes the S value used in the encoding and the quantization matrix 106.

The block combination unit 209 combines the divided blocks output through the IDCT conversion unit (Inverse DCT) 208 to position the original screen.

The interpolation operation unit 210 interpolates the luminance signal Y and the color difference signals RY and BY outputted from the block combination unit 209 and restores the original signal to R, G, and B in the color coordinate conversion unit 211. Convert to a signal.

As described above, according to the present invention, the video compression encoding method and the decoding method for fixing bits according to the present invention provide a DC of a block in which DCT processing is performed on a predetermined coding unit in order to fix an amount of bits efficiently output in a digital video apparatus. The coefficients are encoded according to the order of each block, and then the positional information by position flag is added to all the AC coefficients except the DC coefficient for each block, and the DC information, the positional information, and Data compression efficiency is improved by fixing the output bits by coding data values.

Claims (3)

A video compression coding method for dividing an original video signal into a luminance signal and a chrominance signal, dividing it into blocks having a predetermined size, performing discrete cosine transform (DCT), and quantizing and encoding a DCT coefficient; Encoding a DC coefficient among the plurality of DCT coefficients quantized in the divided block units according to a block order; And encoding the AC coefficient of each of the divided blocks into position data by position flag for each pixel and data values of AC coefficients of pixels having a predetermined absolute value. The video compression coding method for fixing bits. The video compression coding method according to claim 1, wherein the position information is represented by 63 bits that can be represented by 63 position flags other than the DC coefficient with respect to 8x8 blocks. DC coefficients among a plurality of divided quantized DCT coefficients in block units are encoded according to the block order, and the AC coefficient of each block is applied to the AC coefficient of a pixel having a large position information and a predetermined pixel value for each pixel. A video compression decoding method for reconstructing a signal encoded by the data value for the inverse and quantized, and inverse discrete cosine transform the inverse quantization coefficient to output as an original video signal; Decoding the DC coefficients of each block according to the coding order for the divided block units and restoring to a first position of each block; A second step of decoding the AC coefficient to an original value according to the position information of the position flag and restoring it to a corresponding position of the block; And a third process of restoring all AC coefficients without position information in the second process to " 0 ".