JP2908459B2 - Image coding method - Google Patents

Description

[Detailed Description of the Invention] [Object of the Invention] (Industrial application field) The present invention is an image coding method according to a compression coding method for a still image or a moving image.

(Prior Art) When compressing and encoding a still image or a moving image, there is a demand for encoding one screen within a predetermined number of encoding bits as efficiently as possible.

Conventionally, in response to such a request, the number of blocks to be coded in one screen is N for the total number of coded bits B in one screen.
Then, the number of coded bits per block is defined as B / N, and the coding process is performed while adjusting the number of coded bits of each block to a value close to B / (W ·
H.Chen, W.K.Pratt, "Scene Adaptive Coder", IEEE
Trans.COM-32, No.3.1984) and DCT for each block
(Discrete cosine transform), and among the transform coefficients to be transmitted, only the first coefficient counted from the low band is transmitted for all the transmission blocks in one screen.
There has been a method of repeating the operation of transmitting only the third coefficient in the same manner, and aborting the operation when the specified number of bits is reached.

However, none of these methods takes into account the local properties in the screen, and a reproduced image in which the coding error differs depending on the block is obtained.
Alternatively, there is a drawback that the edge of a fine portion of the picture is blurred and the like, which causes a remarkable decrease even in the subjective evaluation.

(Problems to be Solved by the Invention) As described above, conventionally, when one screen is encoded with a predetermined number of encoding bits, the distance is determined without considering local properties in the screen. Due to lack of bit allocation to small parts of the picture, or excessive bit allocation to coarse parts of the picture, etc., the coding error differs depending on the block,
This has led to a decrease in subjective evaluation.

Therefore, the present invention solves these problems, and adaptively allocates coded bits according to local characteristics in a screen, that is, bit allocation in proportion to the degree of fineness of a picture or the existence ratio of high-frequency components. And a condition that coding within one screen within a predetermined number of coding bits is required, and further, a coding error is small depending on blocks, and a good reproduced image can be obtained by subjective evaluation. It is an object of the present invention to provide a simple image coding method.

[Structure of the Invention] (Means for Solving the Problems) In order to achieve the above object, an image encoding method according to the present invention firstly uses a local property within one input screen, that is, a high-frequency component power for each block. Or the standard deviation or variance of the pixel values in the block or the difference between the maximum pixel value and the minimum pixel value of the pixels in the block is obtained from a part or the whole of the screen, and thereafter, one predetermined screen is encoded. The total number of bits for each block in the screen according to the above-mentioned local property (for example, the number of bits is proportional to the variance)
Assign the number of bits to. At the time of actual encoding, the numerical value evaluation (for example, SNR) or the subjective evaluation is discarded so that each block can be encoded with the allocated number of bits, or by discarding those which are not important, or by giving priority to the important ones. By transmitting, each block is encoded with the number of allocated bits, and the entire screen is encoded within a predetermined range of the number of encoded bits.

(Operation) In this way, the predetermined total number of bits for encoding one screen is allocated according to the local property in the screen,
By performing the encoding process of each block based on the assigned bit number, one screen can be encoded with a predetermined encoding bit number, and the encoding error is within the range of the encoding bit number. It is possible to obtain a reproduced image with good subjective evaluation while minimizing the deflection.

(Example) Hereinafter, an example of the present invention is described with reference to drawings.

FIG. 1 is a block diagram according to a first embodiment of the present invention. First, the input screen is divided into predetermined blocks (here, for example, a block of 8 × 8 pixels) by the block dividing circuit 110, and then the amount of generated information for each block (for example, the variance and standard deviation of each pixel value in the block, Alternatively, the generated information amount calculation circuit 111 calculates the difference between the maximum value and the minimum value of the pixel values. Further, in the bit allocation determination circuit 112, the total number of bits that can be used to encode one predetermined screen is proportional to the generated information amount obtained by the generated information amount calculation circuit 111, or the visual characteristics are improved. In such a manner, a portion where the luminance changes gradually changes is weighted and distributed to each block. This allows the number of encoded bits for each block to be, for example, (X: the number of allocated bits of a block, A: the total number of bits that can be used for encoding one screen, N: the sum of the amount of information generated in a block per screen, n: the amount of information generated in a predetermined block). Thereafter, a DCT (Discrete Cosine Transform) circuit 113 applies DCT to each block, and a quantization circuit 114 performs quantization (linear quantization or non-linear quantization).

Here, it is also possible to introduce an adaptive step size method for changing the step size of the quantizer based on the total amount of generated information in one screen calculated by the generated information amount circuit 111 (operated by the dotted line in the figure). is there. However, in this case, it is necessary to transmit the step size information once per screen (or only when the step size is changed several times within one screen). Further, a block signal notifying the end of each block is sequentially input from the block division circuit 110 to the generation information calculation circuit 111. Further, the bit allocation determination circuit 112 may be configured so that the total number of bits (amount of information) that can be used for encoding one screen (one frame) can be manually determined by an external selection means. .

 The configuration of the next stage will be described.

Reference numeral 121 denotes a fixed-length circuit unit, where each D after quantization is
The CT coefficient is divided into a DC component (DC component) and an AC component (AC component), and the DC component is Huffman-encoded by a Huffman encoding circuit 116 by a DC component Huffman table 115 (FIG. 4, step 1). . Further, in this case, instead of encoding the DC component value as it is, a difference value between the DC component of the previous block and the DC component of the previous block may be encoded using Huffman encoding.

On the other hand, regarding the AC component, the Huffman table for AC 118 for the coefficient value quantized for each block and the Huffman table for 0 run 11 for the 0 run (run length).
Huffman coding is performed using 9. At this time, for example, as shown in FIG. 2, when the AC component of the transform coefficient is transmitted by zigzag scan (see IEEE Trans.COM-32, No. 3.1984) from the lower band, the count value to be actually transmitted Is 20,30,50,25,10,1
It is assumed that there is a portion where AC runs of 0 run exist between them. In the fixed-length coding circuit 117, first, these first calculates the bit coefficients N ₁ necessary for assuming all transmitting AC component value and zero-run non-zero (Fig. 4 step 2). If the number of bits N ₁ is X ≧ N ₁ with respect to the coded bit coefficient X assigned to this block determined in advance by the bit allocation determination circuit 112, the coding is performed as it is, but if X <N ₁ (Fourth step 3) includes 0 as shown in FIG.
The absolute value minimum coefficient (10 in this example) other than the above is set to 0 (Step 4 in FIG. 4). Therefore, the AC component values 25 to 15 constitute one zero run, and the AC component having the value of 10 is not transmitted. In this state, re-calculate the encoded bit coefficient N ₂ of this block.

The above operation is repeated, and finally, when the number of encoded bits Ni of this block becomes X ≧ Ni, those coefficients and 0
Each run is transmitted in Huffman code.

Here, as a method of reducing the number of encoded bits Ni of the block to X or less, a method of erasing small AC component values is used, but erasing is performed in order from high-frequency components in consideration of subjective evaluation of reproduced images. Techniques are also conceivable. In addition, instead of the erasure method, a method of transmitting the AC component value in descending order or a method of transmitting the low frequency component in order may be used to stop the transmission at X.

Here, when performing fixed-length encoding for each block, if X> Ni for a certain block (that is, X-Ni bits are left over the actual number of allocated bits), X A method may be employed in which the next block is encoded after adding to the assigned bits of the next block. In other words, the number of bits allocated to the next block is now Xa
Then, from the Ni obtained in step 3 in FIG. 4, Xa + (X
-Ni) is newly set as Xa (step 5), and the processing of one block is ended. In this way, effective bit allocation can be performed within limited bits. Note that the information for each block is separated by transmitting an EOB (end of block) signal. The DC and AC components Huffman-coded by the Huffman coding circuit 116 and the fixed coding circuit 117 are multiplied by the multiplexing circuit 120 and transmitted to the receiving side.

In the processing of the first half of the transmitting side, the order from the block separation circuit 110 to the quantization circuit 114 is first separated into blocks by the block separation circuit 110 and then by the discrete cosine transform by the DCT 113 as shown in FIG. After that, the generated information amount calculation circuit 111 generates all or part of the DCT coefficients (for example, all of the AC components, or the middle frequency portion 701 of the AC components of the AC component block diagram in FIG. 12). Means of calculating the information amount and determining the bit allocation for each block by the bit allocation determining circuit 112 may be considered.

The signal multiplexed and transmitted from the transmitting side is first input to the decoding unit 123, which is separated into an AC component and a DC component by the separating circuit 122 on the receiving side. Here, one DC component is decoded by the Huffman decoding circuit 125 based on the same DC Huffman table 124 as the transmitting side, and the other AC component is the same A as the transmitting side.
Huffman table for C 127,0 Run Huffman table 128
Is decoded by the decoding circuit 126 based on Decrypted DC
The component and the AC component are combined by the combining circuit 129 and the inverse quantization circuit 130
Is inversely quantized. This signal is further subjected to an inverse discrete cosine transform 131, which is the inverse of the conversion on the transmitting side, and each block is frame-synthesized by a frame synthesizing circuit 132, returned to the original signal, and transmitted to a display device (CRT or the like). A signal is displayed. It should be noted that such a transmission / reception device is a system mainly suitable for still image transmission.

The above-described generated information amount calculation circuit 111 shown in FIG. 1 will be described in detail. Here, a method of calculating the amount of generated information for each block based on the difference between the maximum value and the minimum value in the pixel values as described above will be described with reference to FIGS. 13 and 14. FIG. 13 is a block diagram of this circuit, and FIG. 14 is a flowchart. The signals from the preceding block division circuit 110 are compared in order in the pixel value in the in-block pixel value comparison circuit 701 and are arranged in ascending or descending order (step 1).
As soon as the end of the block is notified by the block signal from the block dividing circuit, the maximum pixel value and the minimum pixel value of one block are selected by the circuits 702 and 703, respectively, and their difference a is calculated by the subtraction circuit 704 (Max-Min = a). ) (Step 2). When a of one block is obtained, this value is stored in the memory 705 (step 3). When this operation is performed for each block of one frame and the processing for one frame is completed (step 4), the difference ai (i− = 1,..., N; N) between the maximum pixel value and the minimum pixel value is calculated in operation 706. Is the number of blocks in one frame). Is calculated for each block. (I = 1,..., N) are sent to the subsequent bit allocation determination circuit. Here, the bit bi of each block is assigned ( i = 1,..., N; B is the total number of bits in one frame, and B is variable by the selection circuit.

 Next, another embodiment of the present invention will be described.

FIG. 5 is a block diagram on the transmitting side according to another embodiment of the present invention. An input image is divided into blocks (for example, 8 × 8 or 16 × 16) by a block dividing circuit 301, and a difference (prediction error) from a signal predicted by a predictor 312 is calculated by a subtraction circuit 302. In addition, as processing in the predictor,
Processing such as motion compensation, inter-frame prediction, intra-frame prediction, and prediction using a background memory can be considered. The difference signal is sent to the significant block determination circuit 303, and only a signal having a value equal to or greater than a predetermined threshold (for example, an average luminance value in the block of 4 or more) is transmitted as a significant block. The bit allocation circuit 304 calculates the amount of information in all significant blocks in one screen, and distributes a predetermined number of coded bits for encoding one screen in accordance with the amount of information. To determine the number of coded bits for each block. After that, each significant block is encoded by the encoding circuit 305 (for example, DCT).
After being quantized by the quantizer 306, the fixed length circuit 307
Then, the fixed-length encoding is performed in the same procedure as the processing in the fixed-length encoding circuit 121 in FIG. Thereafter, this signal is inversely quantized by an inverse quantization circuit 308, decoded by a decoding circuit 309, added to a prediction signal from a predictor 312 by an addition circuit 310 to form a reproduction block, and stored in a frame memory 311. Is stored in

When vector quantization (VQ) is used for encoding, the dotted line in FIG. 5 may be changed as shown in FIG. That is, code blocks C ₁ , C ₂ ,…, C _N of different sizes
Using the macro code block 403 prepared a plurality of
A code block of a size represented by the number of bits closest to the number of bits determined by the bit allocation circuit 401 or a codebook of the largest bit in the codebook represented by the number of bits equal to or less than the number of bits determined by the bit allocation circuit 401 A code block is selected from the code blocks 403, and the vector quantization circuit 402 performs vector quantization (V
Perform Q). However, in this case C _1, C _2, ..., it is necessary to transmit or information using any of the codebook C _N to the receiver.

The method described with reference to FIG. 5 is a method mainly used for coding moving images.

Next, another embodiment according to the present invention will be described with reference to FIGS. 7 to 10. FIG. FIG. 7 is a block diagram on the transmitting side according to this embodiment. The block division circuit 501 and the generation information calculation circuit 502 are the same as those described in FIG. When the amount of information (activity) in the block is calculated by the generated information amount calculation circuit 502, signals of this information amount are sequentially stored in the memory 503 for one frame. This signal is also input to a sorting calculation unit 504, a signal for one frame is sorted (step 1), and the maximum value and minimum value are updated and stored in the maximum value storage unit 505 and the minimum value storage unit 506, respectively. . Using these values obtained for each frame, the glass creation unit 507 creates, for example, a class that divides the interval between the maximum value and the minimum value into four. When four classes are defined, the class classifying unit 508 stores the information amount of each block for one frame in the memory 503.
Read from the above four classes, class 1 to class 4
Each block is classified. In this situation, as shown in FIG. 9, each block 521 for one frame is classified into classes 1 to 4.

One of four preset bit allocations is selected by the bit allocation determining circuit 511 according to the glass information (1 to 4) from the class classification unit 508. Based on the bit allocation information (number of bits) from the bit allocation determination circuit 511,
In the encoding circuit 510, the AC component of the information of each block from the class classification unit 508 is encoded. In the Huffman coding circuit 509, the DC component is coded. The encoding is the same as that described in FIG. Such information is multiplied and transmitted by the multiplexing circuit 512 in the same manner as described above.

The bit allocation determination circuit 511 will be described with reference to FIG. Switches 1 to 4 of switches 530 and 533 are set according to the classification information determined by the preceding class classification unit.
For example, when the class i (i = 1... 4) is selected, the bit allocation calculation unit 532 uses the information Xi (= i / 10, (i = 1... 4)) of the bit allocation unit 531 and the bit allocation bi for each block. (B
i = B · Xi / Ni; B is the total number of coded bits in one frame, Ni
(Where i = 1,..., 4) is the number of blocks classified into each class), and the bit allocation bi of each block is output to the encoding circuit in the next stage. The circuits 531 and 532 may be calculated in advance, but this example is effective when the total number of encoded bits B per frame is variable from outside.

As a result, it is possible to perform encoding with the set number of bits, and to obtain a good reproduced image in which the degree of encoding error is minimized within the range of the number of bits.

[Effects of the Invention] As described above, according to the present invention, it is possible to perform encoding processing of one screen within a predetermined range of the number of encoded bits, and to allocate a large number of bits to a portion with a large amount of generated information. Since adaptive bit allocation is performed, it is possible to minimize occurrence of coding error due to local characteristics in a screen, and a good reproduced image can be obtained even by subjective evaluation.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention, and FIG.
FIG. 3 is a diagram showing a state of a conversion surface after performing DCT, and FIG. 3 is a diagram for explaining that absolute value minimum coefficients other than 0 are replaced with 0 by performing fixed length processing. 4
FIG. 5 is a flowchart for explaining an algorithm of the fixed length processing in units of one block, FIGS. 5 and 6 are block diagrams showing another embodiment of the present invention, and FIG. 7 is another embodiment of the present invention. , FIG. 8 is a flowchart for explaining FIG. 7, FIG. 9 is a diagram for explaining the classification, FIG. 10 is a diagram for explaining the configuration of FIG. 11 is a block diagram for explaining a modification of the configuration of FIG. 1, FIG. 12 is a block diagram of a conversion plane, FIG. 13 is a block diagram showing a generated information amount calculation circuit, and FIG. 14 is a block diagram of FIG. It is a flowchart for explaining. 121: fixed length processing unit, 200: conversion surface block, 201: DC component, 202: AC component, 203: coefficient of the smallest value other than 0 among the AC component coefficients, 204: 0 AC component coefficient replaced by 205,... 0 run, 403... Macro codebook with multiple codebooks.

Claims (8)

(57) [Claims] 1. A local characteristic of an image corresponding to a generated code amount of each block of an input image divided into a plurality of blocks is obtained, and a total number of coding bits for coding the input image is determined. In an image coding method in which an image of each block is orthogonally transformed according to the number of coding bits allocated to each block after being allocated to each block according to the distribution of local properties, an image of each block is orthogonally transformed. After that, the generated code amount is allocated to each block by preferentially using the number of coded bits allocated to each block from the larger value of the DC component value and the high frequency component value of the transform coefficient. An image coding method, wherein a transform coefficient is selected so as to be within the number of coding bits obtained. 2. A local characteristic of an image corresponding to a generated code amount for each block of an input image divided into a plurality of blocks, and the total number of coding bits for encoding the input image is determined. In an image coding method in which an image of each block is orthogonally transformed according to the number of coding bits allocated to each block after being allocated to each block according to the distribution of local properties, an image of each block is orthogonally transformed. After that, by using the number of coded bits assigned to the block preferentially from the value of the low-frequency component of each frequency component of the transform coefficient, the generated code amount is assigned to each block. An image coding method characterized in that transform coefficients are selected so as to fall within a number. 3. A local property of an image corresponding to a generated code amount for each block of an input image divided into a plurality of blocks is obtained, and a total number of coding bits for coding the input image is determined. In an image coding method in which an image of each block is orthogonally transformed according to the number of coding bits allocated to each block after being allocated to each block according to the distribution of local properties, an image of each block is orthogonally transformed. After that, by removing the DC component value and the high frequency component value of the transform coefficient in order from the smaller value, the generated code amount is assigned to each block by using the number of coded bits assigned to each block. An image coding method, wherein a transform coefficient is selected so as to be within the number of coding bits obtained. 4. A local property of an image corresponding to a generated code amount for each block of an input image divided into a plurality of blocks, and a total number of coding bits for coding the input image is determined. In an image coding method in which an image of each block is orthogonally transformed according to the number of coding bits allocated to each block after being allocated to each block according to the distribution of local properties, an image of each block is orthogonally transformed. Then, by using the number of coded bits allocated to each block except for the value of the DC component of the transform coefficient and the value of the high-frequency component in order of the value of the high-frequency component, the generated code amount is reduced for each block. An image coding method characterized in that a transform coefficient is selected so as to be within the number of coding bits assigned to. 5. A method according to claim 1, wherein the input image is divided into a plurality of partial images, and predetermined to encode the input image in accordance with local characteristics of the input image corresponding to the degree of fineness of a picture for each partial image. After allocating the total number of coded bits to each partial image, and then orthogonally transforming the partial image in the image coding method of orthogonally transform-encoding the partial image according to the number of coded bits allocated to each partial image, , Which is preferentially selected from the DC component value and the high-frequency component value of the transform coefficient having a larger value, so that the amount of code to be output does not exceed the number of coded bits assigned to each partial image. An image coding method characterized by coding a transform coefficient. 6. A method for dividing an input image into a plurality of partial images and determining the input image in advance in accordance with local characteristics of the input image corresponding to the degree of fineness of a pattern for each partial image. After allocating the total number of coded bits to each partial image, and then orthogonally transforming the partial image in the image coding method of orthogonally transform-encoding the partial image according to the number of coded bits allocated to each partial image, Of the frequency components of the transform coefficients are preferentially selected from the values of the low frequency components, and the transform coefficients are selectively selected so that the amount of code to be output does not exceed the number of coding bits allocated to each partial image. An image encoding method characterized by encoding. 7. An input image is divided into a plurality of partial images, and predetermined to encode the input image according to a local property of the input image corresponding to a degree of fineness of a picture for each partial image. After allocating the total number of coded bits to each partial image, and then orthogonally transforming the partial image in the image coding method of orthogonally transform-encoding the partial image according to the number of coded bits allocated to each partial image, By selecting a transform coefficient by removing the DC component value and the high-frequency component value of the transform coefficient in order from the smallest value, the code amount to be output exceeds the number of encoded bits assigned to each partial image. An image coding method characterized in that transform coefficients are selectively coded so as not to exist. 8. An input image is divided into a plurality of partial images, and predetermined to encode the input image according to a local property of the input image corresponding to a degree of fineness of a picture for each partial image. After allocating the total number of coded bits to each partial image, and then orthogonally transforming the partial image in the image coding method of orthogonally transform-encoding the partial image according to the number of coded bits allocated to each partial image, By selecting the transform coefficient by sequentially removing the DC component value and the high-frequency component value of the transform coefficient from the value of the higher frequency component, the code amount to be output is the number of coded bits assigned to each partial image. An image coding method characterized in that transform coefficients are selectively coded so as not to exceed.