JPH0628444B2 - Image coding method - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to an image coding method for compressing and coding position information and amplitude information of transform coefficients obtained by performing orthogonal transform on an original image.

[Background Art] Conventionally, as an image coding for compressing a data amount by using the statistical property of an image signal, an extrapolation prediction sine transform coding method (Yamane, Morikawa, Hamada: "2D extrapolation prediction-discrete "High-efficiency image coding method using sine transformation", 9th Information Theory and its Application Symposium Proceedings, 1986) is known. As shown in FIG. 4, this image encoding system has a sine transform unit 1 for performing a discrete sine transform, and the block obtained by dividing the original image is decoded from the block of the reproduced image which is decoded after being encoded. After removing the redundancy between blocks by performing the insertion prediction, the redundancy of the prediction error is removed by performing the discrete sine transform on the prediction error. That is, the prediction signal sequence extrapolated by the reproduced image The prediction error sequence {yij} obtained by subtracting each element of the above from each element of the pixel signal sequence {Xij} of the original image is subjected to the discrete sine transformation in the sine transformation unit 1 to obtain the transformation coefficient {Yij}, and It is converted into a compressed code through the conversion coefficient coding unit 2.
Predicted signal sequence Is obtained as follows. That is, after the compression code is decoded into the transform coefficient sequence {ij} in the transform coefficient decoding unit 3 and the prediction error sequence {ij} is obtained in the inverse sine transform unit 4, the prediction error sequence {ij} and the prediction signal sequence And are added, and this value corresponds to the pixel signal string {ij} in the reproduced image block and is stored in the delay memory 5. The image signal sequence {ij} stored in the delay memory 5 is input to the extrapolation prediction unit 6 that performs extrapolative prediction, and the predicted signal sequence Is converted into a pixel signal sequence {Xij} of the original image that is input after that. On the other hand, a reproduced image is obtained from the compressed signal obtained in this way in order to obtain a compressed code. It is possible to apply the same method as that obtained in FIG. 4, and as shown in FIG. 4, the transform coefficient decoding unit 3 decodes the compression code to obtain the transform coefficient sequence {ij}, which is obtained by the inverse sine transform unit. 4, a prediction error sequence {ij} is obtained by inverse sine transformation, and a prediction signal obtained by extrapolatively predicting the pixel signal sequence of the reproduced image stored in the delay memory 5 by the extrapolation prediction unit 6. Row Then, by adding for each component,
ij} can be obtained. In this image coding method, since extrapolation-based prediction is performed to remove correlation (redundancy) between blocks and then redundancy of prediction error is removed, 4 × 4 even if coding distortion is taken into consideration. By using a block of approximately pixels, a relatively small amount of hardware is used, and an excellent encoding characteristic (that is, compression rate of an original image) can be obtained with a relatively small amount of calculation.

As shown in FIG. 6, the transform coefficient coding unit 2 uses Sc
ene Adaptive Coding method (US Pat. No. 4,394,7
74, US Pat. No. 4,302,775, etc.) are used. That is, the transform coefficient sequence {Yi in one block B
j} is scanned in a zigzag pattern as shown in FIG.
After converting the conversion coefficient string {Yij}, which is the dimensional information, into one-dimensional information, the coefficient determination unit 11 compares the preset threshold value T with each component, and the conversion coefficient larger than the threshold value T. Is a significant coefficient and the others are insignificant coefficients. The significant coefficient is quantized by the quantizing unit 13 and then converted into a variable length code corresponding to the quantized value by the variable length coding unit 14. And is output as the amplitude compression code C _A. Further, when the transform coefficient is determined as the insignificant coefficient, the transform coefficient is treated as 0, the run length of the insignificant coefficient is obtained by the counter 15, and the run length value is encoded by the variable length encoding unit 16 to obtain the variable length. A run length compression code C _RL is obtained. In order to identify whether it is the amplitude compression code C _A or the run length compression code C _RL ,
The run length compression code C _RL has a run length prefix code F.
_P is output in front. When the insignificant coefficient continues in the block from the middle to the end, the run length is not encoded and the block end code F _E is output to terminate the encoding of the block. In the block having no significant coefficient, only the block end code F _E is output. The generation of the run-length prefix code F _P and the block end code F _E is managed by the control unit 12. As described above, the amplitude information is compressed as the amplitude compression code C _A and the position information is compressed as the run length compression code C _RL .

By the way, when such processing is performed, the insignificant coefficient is rounded to 0, even though it actually has some value, so that the conversion coefficient string {Yi
If all the components of j} are insignificant variables, all the components of the transform coefficient sequence {ij} which is the output of the inverse sine transform unit 4 are also 0, and in the end, the pixel signal sequence {ij of the reproduced image } Is the predicted signal sequence Will match. However, if there is a region in the original image with a very gradual change in shading, the value of each component of the prediction error string {yij} in this region becomes small, and all the components of the transform coefficient string {Yij} are insignificant coefficients. Certain unintentional blocks are likely to occur. Insignificant block is the prediction signal sequence And the pixel signal sequence {ij} of the reproduced image match, if an insignificant block continuously occurs, the value predicted by extrapolation is used to perform extrapolation prediction, resulting in a prediction error. The gradation level of the reproduced image is gradually accumulated and gradually shifts from the original image. When the prediction error exceeds a certain value, it becomes a significant block including a significant coefficient, so the gray level of the reproduced image approaches the original image here,
This change may appear as a visible distortion because the gray level changes abruptly when a significant block occurs.

[Object of the Invention] The present invention has been made in view of the above points, and an object of the present invention is to reduce the frequency of continuous occurrence of insignificant blocks consisting of only insignificant coefficients. An object of the present invention is to provide an image coding method that suppresses the accumulation of prediction errors due to the insertional prediction and reduces the distortion caused by a sudden change in the gray level of a reproduced image.

DISCLOSURE OF THE INVENTION (Structure) The image coding method according to the present invention divides an original image into blocks having a plurality of pixels, and extrapolates the pixel signal sequence in each block from the blocks coded up to that point. The prediction error sequence is compared with the prediction signal sequence obtained by the prediction, and the prediction error sequence is subjected to orthogonal transformation to obtain a transform coefficient sequence, and each transform coefficient included in the transform coefficient sequence is compared with a threshold value. Among them, the conversion coefficient larger than the threshold value is determined as the significant coefficient, and the conversion coefficient smaller than the threshold value is determined as the insignificant coefficient,
In an image coding method that encodes the position and value within a block only for the significant coefficient, the significant coefficient is included for the block that is located near the block to be encoded and has already been encoded. It is determined whether it is a significant block or an insignificant block consisting of only insignificant coefficients, and the above threshold is set smaller as multiple blocks near the block to be encoded include more insignificant blocks. In this way, the frequency of continuous occurrence of insignificant blocks consisting only of insignificant coefficients is reduced.

(Embodiment) Since the basic configuration of the present invention is the same as the configuration of FIG. 4 and only the transform coefficient coding unit 2 is different from the configuration of FIG. 4, only the transform coefficient coding unit 2 will be described below. Decide to do.

As shown in FIG. 1, a coefficient signal sequence {Yi obtained by orthogonally transforming an error signal sequence which is a difference between a pixel signal sequence corresponding to a block of an original image and a prediction signal sequence predicted by extrapolation
Each component of j} is compared with the threshold value T by the coefficient determination unit 11, and if the value of the component is greater than or equal to the threshold value T, it is determined as a significant coefficient, and if smaller than the threshold value T, it is determined as an insignificant coefficient. The threshold value T is variable based on a specific condition.

The basic principle of setting the threshold T will be described below with reference to FIG. In FIG. 2, the shaded areas indicate blocks that have already been encoded. Now, assuming that the block E is a target to be encoded, blocks B and C adjacent to this block E are referred to. Block B and Block C
For and, the setting criterion is whether or not the components in these blocks B and C include a significant coefficient, and if the block includes a significant coefficient, it is a significant block, and if it is a block consisting of only insignificant coefficients, it is an insignificant block. Called. However, when both blocks B and C are both insignificant blocks, the threshold value T is lowered by a certain amount even when the immediately preceding block (block B in this case) is encoded. If the threshold value T is lowered, the frequency of occurrence of the insignificant coefficient in the coefficient determination unit 11 is reduced, and it is possible to suppress the continuous occurrence of insignificant blocks. On the other hand, if the threshold value T decreases unilaterally by performing such processing, the frequency of occurrence of significant blocks increases and the compression efficiency decreases, so that the blocks B adjacent to the block E to be encoded are When both C are significant blocks,
The threshold value is increased by a certain amount as compared with the case where the immediately preceding block (block B) is encoded. Of course, when only one of the blocks B and C is a significant block, the threshold T is not changed.
In this way, the threshold value T is adjusted according to the blocks B and C adjacent to the block E to be coded. Therefore, if the adjustment amount of the threshold value T is appropriately set, only insignificant blocks will be obtained. , And the compression efficiency does not drop extremely, and the relationship between the degree of distortion and the compression efficiency can be optimized. Also, the block E to be encoded
Even if the threshold value T is adjusted in consideration of not only the blocks B and C adjacent to the block B but also the information of the blocks A and D in the vicinity, almost the same effect can be obtained.

The above adjustment of the threshold value T is performed by using a storage unit 17 and a threshold value setting unit 18 as hardware, as shown in FIG. That is, it is determined through the control unit 12 whether or not a significant coefficient is included in the block. If no significant coefficient is included in the block, the block is stored in the storage unit 17 as an insignificant block, If even one significant coefficient is included, the block is stored in the storage unit 17 as a significant block. Therefore, in the storage unit 17, the position information (indicated by the memory address in the storage unit 17) regarding blocks existing around the block to be encoded and whether the block is a significant block or an insignificant block. The information and are stored. The threshold setting unit 18 reads necessary information from the storage unit 17 according to the position of the block to be encoded, and sets the threshold T of the coefficient determining unit 11 as described above. . Thus, except that the threshold value T is used, the amplitude compression code C _A corresponding to the luminance information and the run length compression code C _RL corresponding to the position information are output as in the conventional example of FIG. Of. When the coding of one block is completed, information as to whether the block is a significant block or a nonsense block is stored in the storage unit 17, and the coding of the next block is performed.

By the way, if information about all the blocks of the original image is stored in the storage unit 17, the storage capacity is increased. Therefore, it is devised to store only necessary information. That is, the original image is divided vertically and horizontally into a matrix to form a large number of blocks, and the block to be encoded is the block E located at the Mth row and the Nth column in FIG. To do. As shown in FIG. 3, the storage unit 17 has a size capable of storing the identification information of blocks for one row, and the information of the Mth row is stored from the address 1 to the address (N-1). The information of the (M-1) th row is stored after the address N. Therefore, A to D shown in FIG.
As shown in FIG. 3 (a), the information of is stored in the storage unit 17 from the address (N-2) to the address (N + 1). In this way, after the block E to be encoded is encoded using the information of the required position, the information of the address N is changed from the information of the block C as shown in FIG. 3 (b). The information of the block E is rewritten so that the next block can be encoded. In this way, by sequentially updating the information, it is possible to sequentially obtain the setting information of the threshold T for the entire screen while using the storage unit 17 that can store the information for one line.

As described above, the present invention divides the original image into blocks having a plurality of pixels, and obtains the pixel signal sequence in each block by extrapolating from the blocks encoded up to that point. The prediction error sequence is subjected to orthogonal transformation to obtain a transform coefficient sequence, each transform coefficient included in the transform coefficient sequence is compared with a threshold value, and the transform coefficient is calculated from the threshold value. Is set as a significant coefficient, a transform coefficient smaller than the threshold is determined as an insignificant coefficient, and the position and value in the block are coded only for the significant coefficient. For blocks that are located near the target block and have already been encoded, determine whether they are significant blocks that include significant coefficients or are insignificant blocks that consist of only insignificant coefficients. , The threshold is set to be smaller as a plurality of blocks in the vicinity of the block to be encoded include more insignificant blocks, and there are many insignificant blocks around the block to be encoded. The lower the threshold for determining whether the coefficient is the insignificant coefficient or the significant coefficient, it is possible to reduce the frequency of continuous occurrence of the insignificant block consisting only of the insignificant coefficient. That is, it has an advantage that the accumulation of the image can be suppressed and the distortion due to the abrupt change of the gray level of the reproduced image can be reduced.

[Brief description of drawings]

FIG. 1 is a block diagram of an essential part showing an embodiment of the present invention, and FIG.
FIG. 3 is a diagram for explaining the principle of the same as above, FIG. 3 is a diagram for explaining the operation of the storage unit used in the same as above, FIG. 4 is a block diagram showing the basic configuration of the image coding system according to the present invention, and FIG. FIG. 6 is a block diagram of a corresponding image reproducing unit, FIG. 6 is a partial block diagram showing a conventional example, and FIG. 7 is an operation explanatory diagram of the same. 1 is a sine transform unit, 2 is a transform coefficient coding unit, 3 is a transform coefficient decoding unit, 4 is an inverse sine transform unit, 5 is a delay memory, 6
Is an extrapolation prediction unit, 11 is a coefficient determination unit, 12 is a control unit, 13
Is a quantizer, 14 is a variable length encoder, 15 is a counter,
Reference numeral 16 is a variable length coding unit, 17 is a storage unit, and 18 is a threshold value setting unit.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Toshitaka Fujii 1048 Kadoma, Kadoma City, Osaka Prefecture Matsushita Electric Works, Ltd. (72) Inventor Hiroshi Hamada 16-4 Takashima New Mansion, Okayama City, Okayama Prefecture (72) Inventor Yoshitaka Morikawa 3-1-1-18 Asahigaoka, Ejiri, Seto-cho, Akaban-gun, Okayama (72) Inventor Nobumoto Yamane 1-3-3 RB, Tsushimanaka, Okayama-shi, Okayama 103 (56) Reference JP 62-195989 (JP, A) JP-A-56-136093 (JP, A) JP-B-49-31808 (JP, B1)

Claims (1)

[Claims] 1. An original image is divided into blocks having a plurality of images, and pixel signals in each block are extrapolated from blocks of a reproduced image obtained by decoding blocks encoded up to that point. Is compared to the prediction signal sequence, the prediction error sequence is orthogonally transformed to obtain a conversion coefficient sequence, and each conversion coefficient included in the conversion coefficient sequence is compared with a threshold value.
Pixel for which the conversion coefficient larger than the threshold value among the conversion coefficients is set as the significant coefficient and the conversion coefficient smaller than the threshold value is determined as the insignificant coefficient, and the position and the value in the block are coded only for the significant coefficient. In the encoding method, whether the block located near the block to be encoded and already encoded is a significant block containing significant coefficients or an insignificant block consisting of only insignificant coefficients An image coding method, wherein the threshold value is set to be smaller as the plurality of blocks near the block to be coded that are determined include more insignificant blocks.