JPS63172584A - Vector encoding system for picture data - Google Patents

Abstract

PURPOSE:To shorten vector encoding processing time by setting the difference quantity of the vector of a processed block and the vector of standard output data corresponding to the code of a block which is near the picture of the block and whose encoding is terminated to below a prescribed threshold, and setting the code of the block which is minimum and nearby to the code of the result of the encoding processing. CONSTITUTION:In a state where the processing of vector encoding is terminated until the vector Xi-1 of i-1 numbered block and the index corresponding to standard output data is obtained as codes (a), (b)..., the encoding of the vector Xi is executed by collating to the code of the block which is near the picture of the vector Xi, and whose processing is terminated. It there are many parts where luminance or color is mostly constant, and the vector of respective blocks do not differ much, such as the background of a picture for example, encoding is executed with collating to the code of the block whose processing is terminated. Thus, the processing time of vector encoding is shortened.

Description

DETAILED DESCRIPTION OF THE INVENTION [Technical Field] The present invention relates to a vector encoding method for image data used in image information transmission or image processing.

[Prior art]

In order to reduce the amount of data to be transmitted during image transmission, the image data to be transmitted is divided into a large number of blocks of, for example, 8 x 8 pixels, and vectors and vectors containing data for each pixel in these blocks are created in advance. Performs matching processing with the vector of each standard output data in a dataset consisting of multiple standard output data, selects the standard output data that has the most similar vector to the vector of the processed block, and selects the standard output data that has the most similar vector to the vector of the processed block. Vector quantization is performed to select the code attached to the data, and the code is transmitted. On the receiving side, standard output data is selected from the data center prepared in advance using the transmitted code, and the standard output data is The block pattern is restored by the vector of the block.

Figure 2 is a diagram conceptually showing a method for encoding such an image. For example, the original image A to be processed is a digital image in which image data such as brightness and color are quantized for each pixel. This original image A is divided into n blocks S 1 , S2 + . . . , S, for each 8×8 pixel, for example.
The image data of each pixel included in each of these blocks is divided into
In order to encode 8×8 pixel data, m standard output data Y1. Y2. ...,Y
1, vector Xi of the block to be processed S and vector Y1 . Y2. . . , Y, the amount of error indicating, for example, the sum of differences in luminance for each pixel is sequentially calculated and the standard output data Y1. Y2...
For each block St, a matching process is performed to select the one with the smallest error amount from the to-be-processed block St, Y.
S2. ...Perform for each of l5II.

In this way, when the standard output data having the vector Y with the smallest error amount from the vector l-X of each block is selected, the index or hector quantization value associated with the standard output data in advance is selected. It is obtained as code C indicating.

However, conventionally, when performing the above-described matching process on the image data of each block, all the standard output data vectors Y1. Since it refers to Y2, ..., Y, it is possible to obtain the same result as the matching process performed earlier because the brightness or color is almost constant, such as the background of an image, and there is almost no difference in the image data of each block. Even when the difference is large, matching processing is performed by referring to all the standard output data, so there is a problem that processing time is wasted.

〔the purpose〕

The present invention focuses on the above-mentioned problems and aims to shorten the time required for processing vector encoding of image data.

〔composition〕

FIG. 1 is a diagram illustrating the present invention in detail, and shows that the vector encoding process as described above has been completed up to the vector X i -1 of the 5-1st block, and the index corresponding to the standard output data is encoded. The state obtained as a, b, c... is shown. As will be explained later in the embodiment, the encoding of the vector X+ is performed with reference to the code of the block whose processing has been completed in the vicinity of the vector X2 on the image. The figure (A) shows the block to be processed S8. A case is shown in which vector encoding of the vector The figure (B) shows block 3 to be processed.
This figure shows a case where vector encoding of vector X8 is performed with reference to the code of the block (block St1.S!ff1) for which the previous two processes of block St24 have been completed.

Such an image may contain many parts, such as the background of the image, where the brightness or color is almost constant and there is almost no difference in the vectors of each block. There is a high possibility that the code obtained by converting into a block that has been processed will match the code of a block that has already been processed. Processing time is reduced.

FIG. 3 is a diagram showing a circuit according to an embodiment of the present invention, in which vectorization is performed using the vicinity of the block to be processed as the neighboring block described with respect to FIG. 1(A) above.

When the vector X1 of the block to be processed is input to the error determination unit l from the input terminal T and latched by the latch circuit 11, the position information of the block to be processed (for example, the block number i
) is input to the neighborhood calculation circuit 32 of the neighborhood selection section 3, and this neighborhood calculation circuit 32 receives the vector X input from the input terminal T.
The position of a block in the vicinity of i (nearby as shown in FIG. 1(A)) is calculated, an access address is output to a line memory 31 (described later) in which the code of the block in the vicinity is stored, and this line Codes C of neighboring blocks are sequentially read out from the memory 31, and a data set unit 5 stores a plurality of standard output data vectors prepared in advance and codes (indexes) corresponding to each standard output data vector. A vector Y of standard output data corresponding to the read code C is read from the memory buffer 33 and stored together with the code C.

In addition, in the case of the vicinity shown in FIG.
It has an area for recording the code of 1), and the code of each block that has been processed is stored sequentially.

The standard output data vector Y and code C corresponding to the neighboring blocks stored in the memory buffer 33 are sequentially outputted to the error determination unit l and latched by the latch circuit 12, and the standard output data latched by the latch circuit 12 is A vector Y of output data and a vector X latched by the latch circuit 11.

The comparator circuit 13 calculates the amount of error between the two and outputs it to the determination circuit 14. This judgment circuit 14 outputs the error amount to the latch circuit 21 of the minimum error detection section 2 when the input error amount is below a predetermined threshold value, and this error amount is used as the above-described error amount of the error judgment section l. It is latched into the latch circuit 21 together with the code output from the latch circuit 12.

Note that the determination circuit 14 uses the vector X of the block to be processed.
If the amount of error between i and the vector Y of the standard output data corresponding to the neighboring block all exceeds the dark value, a start signal is output to the matching processing circuit 4, and the processed block input with all the standard output data is output. The vector matching process is performed.

The error amount and sign that are below a certain threshold latched by the latch circuit 21 of the minimum error detection section 2 are output to the gate 23, and the error amount is also output to the comparator 22, while the launch circuit Reference numeral 24 latches the smallest error amount below the threshold and outputs it to the other input terminal of the comparator 22, and the error amount output from the launch circuit 21 and this latch circuit The comparator 22 compares the amount of error output from the comparator 24 with the error amount output from the comparator 22 . If the error amount input from the latch circuit 21 is smaller, the gate 23 is made conductive by the comparator 22, and the error amount output from the launch circuit 21 is smaller than the one in the vicinity when the error amount is calculated. Latch circuit 2 along with block data code
If the error amount inputted from the latch circuit 21 is larger, the previously latched error amount and sign will be in a state where they are notched.

In this way, when similar processing is completed for the standard output data vectors corresponding to the codes of all the neighboring blocks output from the memory buffer 33, the latch circuit 24 stores the above error amount among the neighboring blocks. The code of the block with the minimum value is latched, and the latched code or the code obtained by the matching process is stored in the line memory 31, and the code is stored in the line memory 31.
The 7 vector data x the code obtained by the most vector encoding process are combined and output to a subsequent circuit (not shown).

Note that when processing the first block,
The matching processing circuit 4 performs matching processing with all standard output data.

For example, as shown in FIG. 1(B), FIG.
), and is a diagram showing a circuit for performing vector encoding processing similar to that described in connection with FIG. This circuit outputs the code of the nearby block to be referenced instead of the neighborhood selection section 3 described above, and vector encoding processing is performed by operating as follows.

That is, the codes sequentially outputted from the latch circuit 24 of the minimum error detection section 2 by processing the vectors of each block are transmitted from the input end of the selector S selected by the select signal to the stage e (shown in FIG. 1(B)). In the above example, each code is sequentially input to the first stage shift register R1 of Re, and is sequentially latched into this e stage shift register which performs a shift operation by applying a clock pulse. When the next vector is processed, each code of the previous 0 blocks is latched in the shift registers R1, R2, . . . , Re.

When the vector XL of the block to be processed is input, a select signal is output to the selector S, and the output end of the final stage shift register Re is connected to the input end of the first stage shift register R1 via the selector S, and each shift The contents of the register are rotationally shifted and held, and the code of the block before the block to be processed is sequentially output from this final stage shift register Re, and the vector of standard output data corresponding to the output code is stored in the data set section. 5 is read out and latched together with the code into the latch circuit 12 of the error determination unit l,
Thereafter, the same process as described with reference to FIG. 3 is performed to perform encoding.

Note that when a vector of one block is being processed, the contents of each shift register are retained by rotational shifting as described above, and the code obtained as described above is stored in the first stage shift register R1. When latched, the contents of the shift register Re at the most extreme stage are erased, and the previous 0 codes of the block to be processed next become newly latched.

〔effect〕

As explained above, according to the present invention, the amount of error between the vector of the block to be processed and the vector of standard output data corresponding to the code of the block whose encoding has been completed in the vicinity of this block on the image is calculated. , the code of the above-mentioned neighboring block whose error amount is the smallest below a certain threshold value is used as the code of the encoding processing result, thereby reducing the time required for vector encoding processing of image data. be able to.

[Brief explanation of the drawing]

Fig. 1 is a diagram explaining the present invention in detail, Fig. 2 is a diagram conceptually showing the encoding process to which the present invention is applied, Fig. 3 is a diagram showing an embodiment of the invention, Fig. 4 2 is a diagram showing an example of a circuit used in an embodiment in which the immediately preceding several blocks are in the vicinity; FIG.

Claims (3)

[Claims] (1) Divide the image into a number of blocks containing a certain number of pixels, and compare the vectors consisting of the data of each pixel included in these blocks with the preset vectors of standard output data for the blocks. In a vector encoding method for image data in which a code corresponding to standard output data having a vector most similar to the vector of the processed block as a result of the comparison is encoded as the code of the processed block, the vector of the processed block and this Find the amount of error between the standard output data vector corresponding to the code of the block whose encoding process has been completed in the neighborhood on the block image, and find the neighborhood where the error amount is the minimum one below a certain threshold. A vector encoding method for image data, characterized in that the code of the block is used as the code of the block to be processed. (2) The vector encoding method for image data according to claim 1, wherein a block adjacent to the block to be processed on the image is a block adjacent to the block on the image. (3) Vector encoding of image data according to claim 1, characterized in that blocks in the vicinity of the block to be processed on the image are several blocks immediately before the block on the image. method.