KR0160616B1 - Digital image compressing method and device - Google Patents

Abstract

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a digital image compression method and apparatus thereof, which compresses and encodes image data of a screen to reduce a bit rate, and forms compressed blocks, which are the minimum units of coding units, and records the compressed data on a recording medium. In a digital video recording / reproducing method of reproducing recorded data, a predetermined number of N × N blocks are divided into one block of uni-sync blocks, zigzag-scanned for each block, and arranged one-dimensionally. Coefficients of the same positions are collected in a block and arranged in ascending order so that the data area of the sync block can separately record the low frequency components and the high frequency components of the luminance signal and the color difference signal to increase the recording efficiency.

Description

Digital image compression method and device

1 is a block diagram of a general digital image compression encoding and decoding apparatus.

2 is a diagram for explaining a Huffman code table proposed by JPEG.

3 is a diagram for explaining a recording format recorded on a tape by a sync block proposed in the present invention.

4 is a block diagram according to an embodiment of a digital image compression device according to the present invention.

5 is a diagram for explaining a macro block.

FIG. 6 is a diagram for explaining a new Huffman code table used in the digital image compression apparatus shown in FIG. 4. FIG.

FIG. 7 is a diagram for explaining an example in which the sync block shown in FIG. 3 is recorded on a tape. FIG.

8 is a diagram for explaining another example of a sink block.

FIG. 9 is a diagram for explaining another example in which the sync block shown in FIG. 8 is recorded on a tape.

* Explanation of symbols for main parts of the drawings

100: zigzag arrangement 101-104,111,112: block memory

105,113: Multiplexer 106,107,114,115: Counter

108,116 buffer memory 120 entropy coding unit

121: New Huffman table

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a digital video compression method and apparatus therefor, the method comprising determining a recording format suitable for recording digital video compressed data on a recording medium, and a digital video compression method suitable for the recording format. It is about.

Recently, remarkable technological advances in various digital fields have expanded the application of digital video, but the limitations in using digitalized video have been studying image compression methods due to the enormous amount of data required to display digital images directly.

In recent years, standardization work has been undertaken by the Joint Photographic Experts Group (JPEG), so the name in standardization is commonly referred to as JPEG, which is a series of continuous-tome still images that contain grayscale and color images. International standard for digital compression.

This JPEG defines a core mode of operation called the base-line system, which is used to implement other modes of operation, while in many applications, the baseline mode alone is useful.

The digital video compression encoding and decoding apparatus using the encoding and decoding method using DCT used in such JPEG is as shown in FIG.

First, an analog image is converted into digital input and then divided into 8 × 8 pixel blocks. If the size of one screen is 480 (vertical) x 720 (horizontal), there are 5400 blocks.

The DCT converter 10 performs DCT conversion on a divided 8 × 8 block basis. The 8 × 8 image block is a 64 pixel discrete signal that is a function of two spatial dimensions (that is, the x and y axes). The forward discrete cosine transform (FDCT) takes these discrete signals as inputs and separates them into 64 quadrature basis signals. Each orthogonal reference component (ie, DCT coefficient) represents each spatial frequency component when the frequency component of the input signal is represented by 64 two-dimensional spatial frequency components. Therefore, DCT coefficients can be regarded as a representation of unique frequency components determined according to the 64 pixel input image.

The DCT coefficients at x = 0 and y = 0 are called DC coefficients because they represent zero frequency in both dimensions. The remaining 63 coefficients are called AC coefficients. Pixel values within an 8x8 block usually do not change rapidly, so most of the energy is collected in low frequency components. Thus, most of the spatial frequency components are zero or small enough to be nearly zero in size, so there is no need to encode them.

In the quantization unit 20, 64 DCT coefficients, which are outputs of the FDCT, are subjected to linear quantization by the quantization table 30 having 64 values, respectively. Quantization of the DCT coefficients is achieved by dividing the result by the quantization size and then substituting the result for the nearest integer.

After quantization, the DC (DC) coefficients are separated from the 63 AC coefficients and processed independently. Since the DC coefficient is an average value of 64 pixels in a block, it is generally correlated with the DC values of adjacent blocks, so it is more efficient to separate and encode the difference from each other.

In addition, the remaining 63 AC coefficients are rearranged in a zigzag fashion and non-zero AC coefficients distributed in the low frequency region can be seen together and then increase in compression through entropy encoding.

The entropy encoding unit 40 obtains by using statistical characteristics of the quantized DCT coefficients in the encoding. That is, a short codeword is given to coefficient values that appear statistically frequently, and a long codeword is provided in rare cases to reduce the average amount of data required to express coefficients.

Then, in order for the variable length coded code encoded with a binary code to be recorded on a tape, an error correction code is added and then recorded or transmitted to a recording medium (not shown) through digital modulation or the like.

Repeat this process for all blocks. In the process of quantizing the DCT coefficient and expressing it as a variable length code, the amount of data is greatly reduced to obtain a compression effect.

The entropy decoding unit 60 decodes the coded data through the Huffman table 70, and the weakening unit 80 performs inverse quantization using the quantization table 90.

In the IDCT 100, FDCT reverses a process, and 64 spatial domain pixels are reproduced by taking and converting 64 DCT coefficients. Mathematically, since DCT performs a 1: 1 conversion, the original 64 image samples can be reproduced.

Here, the variable length code is encoded by applying the Huffman code used in the baseline system, which is a mode of the above-described JPEG (International Standard for Still Color Image) coding scheme.

At this time, the format of the Huffman code used in the baseline system is shown in FIG.

NNNN is a continuous length of zero values, SSSS is a class value that categorizes non-zero coefficients by category, and NNNN = 0 when SSSS = 0 is EOB (signal indicating end of block), NNNN = 1-14 is used in extended mode, and NNNN = 15 is used when the length of continuous zero is 16 or more.

This Huffman code is shown in FIG. 2 in 1/0, 2/0,... No sign is assigned to the 14/0 position.

Disclosure of Invention An object of the present invention is a digital image compression method for increasing recording efficiency of a recording medium by separately recording low and high frequency components of a luminance signal and a color signal in a data area of a sink block in an image compression apparatus recorded on a recording medium in sync block units. To provide.

Another object of the present invention is to provide a digital image compression apparatus capable of improving recording efficiency of a recording medium by assigning a code to a portion of the Huffman code which is used for variable length encoding in the baseline system proposed by JPEG. To provide.

In order to achieve the above object, the digital image compression method according to the present invention compresses and encodes the image data of the screen to reduce the bit rate, and forms a sync block that is the minimum unit of the coding unit to record the compressed data to the recording medium, A digital image compression method for reproducing data recorded from a recording medium, comprising dividing a predetermined number of N × N blocks into one sync block as one unit, zigzag scanning for each block, and aligning one-dimensionally. Variable length coding is performed by collecting coefficients of the same position for each block and arranging them in ascending order.

In addition, the digital image compression device according to the present invention comprises: discrete cosine converting means for discrete cosine transforming an input video signal; Quantization means for quantizing the transform coefficient output from the discrete cosine transform means; The quantized data output from the quantization means is divided into a sync block having a predetermined number of N × N blocks as one unit, and zigzag-scanned for each block, and arranged one-dimensionally. Rearrangement means for collecting coefficients of position and arranging them in ascending order; And variable length encoding means for variable length encoding the data outputted from the rearrangement means and transmitting the data to the recording medium.

Hereinafter, with reference to the accompanying drawings will be described a preferred embodiment of a digital image compression method and apparatus according to the present invention.

In order to store a variable length code on a tape in an image compression apparatus, a tape recording format must be determined.

The present invention proposes a tape recording format recorded in a sync block as shown in FIG.

In addition, the tape format as shown in FIG. 3 has a new variable length for efficiently recording data in the IDC (Independently Decorable Code) and DDC (Dependently Decorable Code) areas. We want to use a sign table (Huffman table).

This will be described in detail with reference to a block diagram according to an embodiment of the digital image compression device according to the present invention shown in FIG.

First, encoding of quantized data output from a quantization unit (not shown) is usually performed independently by sub-picture units having a constant size. Each block of 8x8 pixels or four luminance blocks and one chrominance signal block are referred to as one macro block or can be encoded independently for every two macro blocks.

One independent coding unit is referred to herein as a sync unit. In the present invention, one macroblock is one sync unit.

One macro block is four luminance signal blurs (Y1, Y2, Y3, Y4), one RY color difference signal block Cr block, and one BY color difference signal block as shown in FIG. It means a subimage composed of Cb blocks.

These blocks are already quantized integer DCT coefficients. In this example, one block is expressed as a block of 4 × 4 size rather than 8 × 8, but the device shown in FIG. 4 can be used as an 8 × 8 block.

In the zigzag arrangement unit 100, zigzag scanning is performed for each block shown in FIG.

1st luminance block Yl: 10 -2 3 0 1 0 0 1 EOB-(1)

2nd luminance block Y2: 11 2 -1 0 -2 0 0 0 0 0 0-1 EOB-(2)

3rd luminance block Y3: 8 0 3 1 -1 EOB-(3)

4th luminance block Y4: 11 1 EOB-(4)

R-Y color difference signal block Cr: -4 1 EOB-(5)

B-Y chrominance signal block Cb: 2 -1 EOB-(6)

In order to combine the blocks of this one-dimensional array into an array, the coefficients of the same position are collected and arranged in ascending order.

The reconstruction of four luminance signal blocks into one one-dimensional array is as follows.

That is, one-dimensional zigzag-arranged data of four luminance signal blocks Y1, Y2, Y3, and Y4 is stored in the first to fourth block memories 101 to 104, as shown in (1) to (4) above. do. The first counter 106 is a four-counter counter that counts from 0 to 3, and this output is input to the first and second selection terminals SEL0 and SEL1 of the first multiplexer 105.

Each time counting 3 in the first counter 106, this output is inputted as a clock signal of the second counter 107, which is a 64-bit counter that counts from 0 to 63 through the NAND gate G1. The output of 107 is incremented by one and input as the read address signals A0-A5 of the first to fourth block memories 101-104.

The first multiplexer 105 selects one of the outputs of the first to fourth block memories 101 to 104 stored at an address output from the second counter 107 according to the output selection signal of the first counter 106. 1 is stored in the buffer memory 108.

The two color difference blocks are reconstructed as one-dimensional arrays as follows.

In the fifth and sixth block memories 111-112, one-dimensional zigzag-arranged data of two color difference signal blocks Cr and Cb is stored as shown in (5) to (6). In 114), it is a binary counter that counts 0 and 1, and this output is input to the first and second selection terminals SEL of the second multiplexer 113.

Whenever the third counter 114 counts 1, the output is inputted as a clock signal of the fourth counter 115, which is a 64-bit counter that counts from 0 to 63 through the inverter INV1. The output of 115 increases by 1 and is input as the read address signals A0-A5 of the fifth and sixth block memories 111-112.

The second multiplexer 113 selects an output of the fifth or sixth block memories 111-112 stored at an address output from the fourth counter 115 according to the output selection signal of the third counter 114, and selects the first output. Stored in the buffer memory 116.

The order of data stored in the first buffer memory 108 is as follows (7).

Luminance signal: 10 11 8 11 -2 2 0 1 3 -1 3 EOB 0 0 1 1 -2 -1 0 0 EOB 0 0 1 0 EOB 0 0 0 -1 EOB-(7)

The order of data stored in the second buffer memory 116 is as follows (8).

Color Signal: -4 2 1 -1 EOB EOB-(8)

In the case of the luminance signal, since there are four blocks, there are four EOBs. Whenever an EOB occurs, the zero run length, which is the number of consecutive zeros before it, is calculated.

In this way, the data stored in the first buffer memory 108 and the second buffer memory 116 is encoded by the entropy encoding unit 120 using the new Huffman table 121 proposed in the present invention. This will be described in detail as follows.

As shown in FIG. 6, in the new Huffman table 121, EOB means SSSS = 0. Therefore, the first EOB can be marked as 2/0 because the zero run length is 2. The remaining EOBs are as follows.

In case of luminance signal

first EOB: 0/0-(9)

second EOB: 2/0-(10)

third EOB: 1/0-(11)

fourth EOB: 1/0-(12)

In case of color signal

first EOB: 0/0-(13)

second EOB: 0/0-(14)

If the end value of the one-dimensional coefficient array of a block is a non-zero coefficient, the corresponding EOB is not displayed.

FIG. 6 shows the new Huffman table 121 shown in FIG. 6, which is not used in the baseline system but used for other purposes in the table shown in FIG. In addition, the values 14/0 can be used to represent zero run-length / EOB (= 0).

Here, 14/0 has two meanings. In other words, EOB occurs after 14 consecutive zeros and EOB occurs after 15 consecutive zeros. One additional bit is added to distinguish this.

The reason for using the above-described DCT coefficient array and the new Huffman table is that it is advantageous to restore and display only data of a constant low frequency component during track reproduction such as high-speed search of DVCR.

Therefore, N macroblocks composed of 4 x N luminance blocks and 2 x N chrominance blocks are referred to as one sync unit, and one-dimensional arrays of luminance and color difference signals are performed. The problem of multiple EOBs in one-dimensional arrays can be represented by the new Huffman code table described above.

Next, a description will be given of one sync block in the tape recording format when the encoded data output from the entropy encoding unit 120 is recorded on the tape according to the recording format proposed by the present invention.

First, DDC is subdecoding code that is decoded only when IDC is decoded and cannot be decoded by itself. Record the IDC at regular intervals for trick play. One macroblock corresponds to lIDC. One frame has the effect of fixing bits in units of 1350 macroblocks.

In the IDC region, the DC coefficients of Y and C and the AC coefficients corresponding to the constant frequency region from the low frequency region are recorded.

This IDC region includes a first independent decodeable code (IDC1) region in which the DC coefficient of Y and the low frequency region coefficients of Y are recorded, and a second independent decodeable code (IDC2) in which the DC coefficient of C and the low frequency region coefficients of C are recorded. Area). The remaining Y and C AC coefficients are recorded in the IDC area. If the screen data of a certain size cannot be recorded in this area, it can be recorded in the next DDC area of the sync block. In addition, S is a resync code, which is ffff when expressed in hexadecimal, and is the number of sync blocks.

On the other hand, the DC components of the Mth sink unit and a certain amount of low frequency components must be recorded in the IDC region of the Mth sink block. The data to be recorded in the IDC area of the Mth sink block need not be data in the enhancement Mth sink unit.

In summary, the sync block includes a sync code (S), an identification code (ID) indicating the number of the sync block, a first independent decodable code (IDC1) independently decodable on the DC coefficient of Y and the low frequency region coefficients of Y, Additional code (P) for error correction of the second independent decodeable code (IDC2) and the first and second independent decodeable codes (IDC1, IDC2) independently decoded for the DC coefficient of C and the low frequency region coefficients of C. ), The sub-decoding code (DDC) which can be decoded dependently on the first and second independent decodeable codes (IDC1, IDC2), and the additional code (P) for error correction of the sub-decoding code (DDC).

If four sync units encoded using the Huffman code, which is a variable length code, have the following data lengths, four sync blocks corresponding thereto are shown in FIG.

Sync Unit 1 = Y1 (800bits) + C1 (300bits)

Sync Unit2 = Y2 (750bits) + C2 (200bits)

Sync Unit 3 = Y3 (300bits) + C3 (70bits)

Sync Unit 4 = Y4 (850bits) + C4 (250bits)

Here, the IDC1 area is allocated 400 bits, the IDC2 area 100 bits, and the DDC area 550 bits.

As shown in FIG. 7, the recording of the luminance signal and the color difference signal must be separately arranged in the IDC of each sync block (IDC1, IDC2), but recording is difficult. That is, as shown in FIG. 8, the DC and AC components of the luminance component are first recorded in the IDC region. The DC and AC of the chrominance signal are pushed to the DDC and recorded when there is no spare to record.

In summary, the sync block includes a sync code S, an identification code ID, an independently decodeable IDC, an additional code P for error correction of the independent decoding code, and the independent decoding code. Dependent decodable code that can be decoded dependently on the subcarrier, and additional code (P) for error correction of the subdecodable code.

This data recording format is simpler than that of Fig. 7, but is disadvantageous for the restoration of color signals during trick reproduction.

500 bits are allocated for the IDC area in the sink block and 550 bits for the DDC area. Although the lengths of the IDC and DDC are different from those described in FIG. 6, the data is written in the IDC area as described above in the order of DC coefficient of Y. , AC coefficient of Y, DC coefficient of C, AC coefficient of C. This IDC area records the DC coefficients of Y and C and the AC coefficients corresponding to the constant frequency range from the low frequency area. The remaining Y and C coefficients recorded in the IDC area are recorded. If the remaining data cannot be recorded in the screen data area of a predetermined size, they may be recorded in the DDC area of the next sync block.

As described above, in the digital image compression method and apparatus according to the present invention, in a digital imaging apparatus such as a D-VCR which is recorded on a recording medium in units of sync blocks, the data area of the sync block is divided into a luminance signal and a color difference signal. The recording efficiency can be increased by separately recording the low frequency component and the high frequency component, and the recording efficiency can be improved by utilizing the continuous zero count / EOB region, which is an area not partially encoded by the Huffman code, for a variable length encoder suitable for this recording format.

Claims (8)

In a digital video recording and reproducing method of compressing and encoding image data of a screen to reduce a bit rate, and forming the compressed data into a recording block by minimizing a coding unit, and reproducing the recorded data from the recording medium. Divide the N × N blocks of integers into a sync block, which is a unit, zigzag scan each block, and arrange them in one-dimensional order.The coefficients of the same position are collected and arranged in ascending order for each one-dimensional block. Digital video compression method for long coding. The digital image compression method of claim 1, wherein the variable length encoding is performed by using all regions of the Huffman table used in the variable length encoding process. 2. The independent independent decodable code according to claim 1, wherein the sync block is independently decoded with respect to a sync code, an identification code representing a representation of the sync block, a DC coefficient of the luminance signal Y, and a low frequency region AC coefficient of Y. (IDC1), addition for error correction of the second independent decodeable code (IDC2), the first and second independent decodeable codes independently decoded with respect to the DC coefficient of the color signal C and the low frequency region AC coefficients of C A code, a subdecodable code that can be decoded dependently on a first and a second independent decodable code, and an additional code for error correction of the subdecodable code. 4. The DC component of the Mth sink unit and a predetermined amount of low frequency components are necessarily recorded in the IDC1 and IDC2 regions of the Mth sink block, and the DDC region is recorded in the IDC1 and IDC2 regions. The AC coefficients of the digital image compression method is characterized in that if a certain size of screen data is not recorded in this area, it is recorded in the next DDC area of the sync block. The method of claim 1, wherein the sync code is a sync code, an identification code, a first data code that can be independently decoded, a first additional code for error correction of the first data code, and can be decoded dependently on the first data code. And a second additional code for error correction of the second data code and the second data code. Discrete cosine converting means for converting the input video signal into discrete cosine; Quantization means for quantizing the transform coefficient output from the discrete cosine transform means; The quantized data output from the quantization means is divided into a sync block having a predetermined number of N × N blocks as one unit, and zigzag scanning is performed for each block, and arranged one-dimensionally. Rearrangement means for collecting coefficients of position and arranging them in ascending order; Entropy that entropy-encodes the data output from the rearrangement means using a new Hoffman table to properly record the independent decodable code, the dependent decodable code, and additional codes for error correction of the codes in a tape recording format. Digital video compression apparatus comprising the encoding means. 7. The apparatus of claim 6, wherein the rearrangement means comprises: zigzag arrangement means for zigzag scanning one block of data of sync blocks of the predetermined number of NxN blocks quantized by the quantization means; A plurality of luminance signal block memories for temporarily storing zigzag arranged outputs of a plurality of luminance signal blocks with respect to the sync block output from the zigzag arrangement means; First selecting means for selecting coefficients of the same position from the outputs of the plurality of luminance signal block memories in ascending order of low frequency to high frequency coefficient; Second selection means for controlling to select coefficients of the same position from the outputs of the plurality of color signal block memories in ascending order of low frequency to high frequency coefficient, and storing coefficients of the luminance signal for the sync block selected by the first selection means; A first buffer memory which is then output to the variable length encoding means; And a second buffer memory for storing the color signal for the sync block selected by the second selecting means and outputting the color signal to the variable length encoding means. 8. The apparatus according to claim 7, wherein said first selecting means rounds up at a coefficient at the same position from switching means for selecting signals output from said plurality of luminance signal block memories, and quantized coefficients zigzag arranged from said plurality of luminance signal block memories; A first address generating means for generating a read address of said plurality of luminance signal block memories so as to be sequentially read out, said second selecting means being a switching means for selecting signals outputted from said plurality of color signal block memories; And a second address generating means for generating a read address of said plurality of color signal block memories to be read out in ascending order from the quantized coefficients zigzag arranged in said color signal block memory of said plurality of color signal block memories. .