JPH02292962A - Block encoding system - Google Patents

Abstract

PURPOSE:To encode a mean value, a difference value, and the key level of the picture elements in a block based on each fixed value by storing the emerging frequency of the different between the maximum and minimum levels of the picture elements in a block when the picture signals are encoded at every block and preparing the block identifiers in accordance with the block modes divided in an optimum manner based on the stored difference emerging frequency. CONSTITUTION:When an encoding process is commanded via the CPU 21 of a block encoding system with the operations of keys, etc., the prescribed number of picture elements in the first block of a picture to be processed are taken out of an original picture storage 22. Then the mean value X, the maximum value MAX, the minimum value MINL, and (MAXL-MINL)=D are calculated respectively for the picture elements in the taken-out block according to an encoding table 23 and the information on a RAM 24. The value D is compared with a parameter TI whether D<=TI is satisfied or not is judged. Then the next process is carried out when the value D is equal to or smaller than the TI. Then the picture elements in a block are expressed in a single level and the process of a relevant mode is carried out.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a block encoding method for compressing television images, facsimile images, etc. for transmission or recording.

[Conventional technology and issues to be solved] Conventionally, when transmitting and recording television images or facsimile images, the images were divided into blocks in order to improve transmission efficiency and recording efficiency, and each block was encoded based on an average value, etc. The image signal is compressed. That is,
As shown in FIG. 4, the image information of one screen 11 is divided into NXN pixels, for example, a block 12a of 4×4 pixels. 12b,...
・The maximum pixel value in each block (MAIL
) and the minimum value (MINL) into four gradations and encoded in bit-brain format. Image information is divided into pixel average value La, difference value Ld, and bit-brain information ψ[. Divided into three components of φ2. Furthermore, the above image information is converted into a parameter (threshold value) TI. It is divided into three modes A, B, and C and expressed using T2 (T1≦72).

That is, when rMAXL-MINL≦TIJ, it is expressed as mode A at one level. At this time, bit brain information φl. is provided one bit for each pixel value. φ2
are all “0°”.

When rTl<MAXL-MINL≦T2J, it is expressed as mode B in two levels. At this time, all bit brain information φ2 is set to “0”, and φ1 has the pixel average value La.
For pixels with larger pixel values, “0°, pixel average value L
A pixel with a pixel value smaller than a is assigned "1°."

When rMAXL-MINL>≦T2 J (7), it is expressed as mode C in four levels. At this time, bit plane information φ1, which is provided one bit for each pixel value,
φ2 is all set to 0°.

In addition, the pixel average 1i1La, the difference 1Ld, and the bit brains φl and φ2 in each of the above modes are shown in FIGS.
It is determined by the process shown in (C).

In addition, in FIG. 5 above, X: Average value of pixels within the block (φl)++: Bitbrain information 1 (1≦l, j≦4) (φ2) 1: Bitbrain information 2 (1≦1, j≦4) X+): Pixel value within the block (1≦l, j≦4) It is.

Then, the pixel average value La and the difference value Ld are further quantized and then encoded using a human code table. Also, bitp1non information φ1. φ2 is encoded using a two-dimensional standard encoding method.

As described above, the pixel average value La in each block,
The difference value Ld and bit plane information φ1, φ2 are encoded and recorded or transmitted.

[Problems to be Solved by the Invention] By using the block encoding method described above, it is possible to encode image information in units of blocks and compress image signals.

However, in the conventional block encoding method described above, when setting an upper limit on the amount of code, the entire process has to be executed multiple times by adjusting the parameters TI and T2, resulting in a problem that the processing takes time. Ta.

The present invention has been made in view of the above circumstances, and an object of the present invention is to provide a block encoding method that can easily perform processing to keep the code amount constant when an upper limit is set on the code amount. shall be.

[Means and effects for solving the problem] In order to achieve the above object, the present invention stores the number of occurrences of the difference between the maximum value and the minimum value of the pixels in the block when encoding the image signal by dividing it into blocks. , based on this, the blocks are divided into optimal block modes, a block identifier corresponding to the mode is attached, and the average value, difference value, and representative level of pixels within the block are encoded.

As described above, by determining the optimal block mode based on the number of occurrences of the difference between the maximum and minimum values of pixels within a block, and adding a block identifier indicating the mode of the block, the amount of code can be kept constant. , and the encoding process can be simplified.

[Embodiments of the invention] Hereinafter, one embodiment of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing the circuit configuration of the present invention. In the figure, 21 is a CPU that controls the operation of the entire system.
An original image storage device 22, an encoding table 23, a RAM 24, and a code word storage device 25 are connected to this CPU 21.

The original image storage device 22 digitally stores each pixel value of an image to be encoded.

The encoding table 23 is used by the CPU when encoding data.
21, and the block identifier encoding table 23a. La encoding table 23b, Ld encoding table 23C, φ1 encoding table 23d.
It consists of a φ2 encoding table 23e. The block identifier encoding table 23a has a 1-bit identifier "0" in mode A, and a 2-bit identifier in mode B.
A bit identifier "10" is assigned, and in the case of mode C, a 2-bit identifier "11" is assigned. The La encoding table 23b is a Huffman codeword table assigned according to the appearance frequency of the difference value between the pixel average value La of the previous block and the current block.
The encoding table 23c for d is a Huffman codeword table allocated according to the frequency of appearance of the difference value Ld. φ
1 encoding table 23d statistically examines the number of occurrences of 16-bit patterns in a block, and selects 1 with a high probability of occurrence.
For 92 patterns, the 8-bit code word “00~BF(
A decimal number 191)J is assigned, 2 bits of "11" are added to the beginning of the remaining patterns, and a code word in which the 16-bit patterns are subsequently arranged as is is assigned. φ2
The encoding table 23e for φl is the encoding table 23 for φl.
Code words are assigned in the same manner as in d.

The RAM 24 is mainly used to temporarily store code words of images to be processed, and includes a block identifier storage section 24a,
La storage section 24b, Ld storage section 24C, φl storage section 24
d, φ2 storage section 24e, and a D storage section 24f, and the D storage section 24f stores D (
This is for counting the number of appearances for each value (D-MAXL-MINL).

The code word storage device 25 includes a block identifier storage section 25a,
L'a storage section 25b, Ld storage section 25C, φ1 storage section 2
5d. It consists of a φ2 storage section 25e.

FIG. 2 shows a functional block diagram of the CPU 21. The image data stored in the original image storage device 22 is read out in units of pixel values for one block by the block reading section 3l, and is stored in the block memory 32. The pixel values for one block stored in the block memory 32 are sent to the maximum value/minimum value detection section 38, the average value detection section 34, the difference value detection section 35, the φ1 detection section 3B, and the φ2 detection section 37. The maximum value/minimum value detection section 33 is
Detects the maximum value and minimum value from the pixel values of one block stored in the block memory 32, average value detection section 34, difference value detection section 35, mode detection section 38, LM maximum value - minimum value) aggregation section 39 Output to. The average value detection unit 34 detects an average value (median value) La from the maximum value/minimum value detected by the maximum value/minimum value detection unit 33 and sends it to the average value encoding unit 4l and the φ1 detection unit 3B. Output. Average value encoding unit 4
l refers to the La encoding table 23b, encodes the average value La, and stores it in the La storage section 24b. The pixel average value La stored in the La storage section 24b is sent to the code remaining amount detection section 42.

Further, the mode detecting section 38 determines the encoding mode from the maximum value/minimum value detected by the maximum value/minimum value detecting section 33, and the average m detecting section 34, the difference value detecting section 35 and the mode identifier storage section Output to 24a. The difference value detection section 35
detects the difference value Ld using the block data from the block memory 32, the maximum value/minimum value from the maximum value/minimum value detection section 33, and the encoding mode from the mode detection section 38, It is output to the φl detection section 3B.

The difference value encoding unit 43 encodes the difference value Ld with reference to the Ld encoding table 23c, and stores the difference value Ld in the Ld storage unit 24c.
l. :Remember.

The difference value Ld stored in the Ld storage section 24c is sent to the remaining code amount detection section 42.

The φl detection unit 3G determines the value of the bit brain information φl based on the information sent from the block memory 82, the mode detection unit 38, the average value detection unit 34, and the difference value detection unit 35, and sets the value to φ1. ? It is output to the f-encoding section 44.

This φl encoding unit 44 has a φ2 encoding table 23d.
The above bit plane information φl is encoded with reference to φ
1 is stored in the storage section 24d. And this φl storage unit 2
4d is input to the code remaining amount detection section 42.

Further, a block identifier is manually inputted to the remaining code amount detection section 42 from the block identifier storage section 24a. The code remaining amount detection unit 42 uses block identifiers, La, Ld, φ1 stored in each storage unit 24a to 24d of the RAM 24.
The remaining code amount is detected by calculating the total code amount, subtracting it from the upper limit value of the code amount, and outputting it to the threshold determination unit 45. Further, the threshold value determination section 45 is given aggregated data from the (maximum value - minimum value) aggregation section 39 . Threshold determination unit 45
is the code remaining amount detection unit 42 and (maximum value - minimum value) collection 1
A threshold value is determined based on the data from section 39 and output to mode detection section 46 . In addition, this mode detection section 46
In this case, the (maximum value - minimum value) tally data of the tally section 39 is manually inputted, and the block identifier stored in the block identifier storage section 24a is read out by the block identifier reading section 47.
Input via . This block identifier reading section 47
determines the mode of the block based on the data from the threshold determining section 45, (maximum value - minimum value) aggregation section 39, and block identifier reading section 47, outputs it to the block identifier storage section 24a, and outputs it to the block identifier storage section 24a. It is output to the detection unit 37. This φ2 detection unit 37 detects bit brain information φ2 in the block according to the mode detected by the mode detection unit 4B, and outputs it to the φ2 encoding unit 48.

This φ2 encoding unit 48 has a φ2 encoding table 23e.
φ2 is encoded with reference to φ2 and stored in the φ2 storage unit 24e. After finishing the encoding process for all blocks of one screen, the block identifier storage unit 24 a s
La storage section 24b, Ld storage section 24c, φl storage section 2
The encoded data Fl stored in the 4d1φ2 storage section 24e is transferred to the codeword storage device 25 in FIG.

Next, the operation of the encoding process in the above embodiment will be explained with reference to the flowchart of FIG. The flowchart shown in FIG. 3 is executed under the control of the CPU 21. When the CPU 21 is instructed to perform encoding processing by a key operation or the like, first, the CPU 21 takes out 4×4 pixels (16 pixels) of the first block of the image to be encoded from the original image storage device 22 (step Sl). Then, the average value X, maximum value MAIL, minimum value MINL, MAIL-M of pixels in the block
Each INL-D is calculated (step S2). Then, compare the value of D and the parameter TI, and find that D≦TI
If the value of D is equal to or smaller than Tl, the process proceeds to step S4; otherwise, the process proceeds to step S7.

Steps 84 to S6 described above are mode A processing in which pixels within a block are expressed at one level. In this mode A process, first, the average pixel value in the block is set to X (step S4), and "O" for mode A is used as the block identifier.
1 bit, block identifier storage section 24 in RAM 24
Store in a. Then, the quantized pixel average value La of the previous block is subtracted from the value of the pixel average value La, and La
The difference is quantized and encoded using the encoding table 23b for d
Remember yourself. (Step SI3).

Further, if the condition of step S3 is not satisfied, mode B or mode C processing is executed. Here, since the value of parameter T2 has not yet been determined, processing common to these modes will be performed. First, a counter for counting the number of appearances prepared for each value of D is incremented by "+1" (step S7). This counter is stored in the D storage unit 24 [
It is located inside. Then, among the 16 pixels in the block, the average value of the pixels that is equal to or larger than r (MAXL-MINL)/2J l is determined and stored as Pi. Also, among the 16 pixels in the block, r (M
The average value of pixels smaller than AXL-MINL)/2J is determined and stored as P2. Then, the calculation La=(Pi+P2)/2Ld-(PI-P2) is performed to obtain the pixel average value La and the difference value Ld. In addition, (φl),, respectively indicate the level of the pixel value corresponding to x1, and Xz(1≦i≦4.1≦j≦
4) for pixels whose value is smaller than the pixel iP average value La (
1 bit “1°” is assigned to φl)z, and 1 bit “0” is assigned to (φl)z for pixels with other pixel values.
(Step S8).

Thereafter, the difference ΔLa between the pixel average field La of the current block and the quantized pixel average v1La of the previous block is quantized and encoded using the La encoding table 23b, and La
The data is stored in the storage unit 24b. Also, the difference value Ld is L d
JTJ? '1 Quantization and encoding are performed using the encoding table 23c, and the results are stored in the Ld storage unit 24c. Furthermore, 16 bits of bit brain information φ1 are encoded in φl encoding table 2.
3d and stored in the ψl storage unit 24d. At this time, for example, the mode B identifier "10" is set as a temporary block identifier in the block identifier storage unit 24al. : Store (step S9).

After performing the processing in step S8 or step S9 as described above, it is checked whether or not processing has been completed for all blocks of the image to be processed (step S10).
). If the processing has not been completed, the process proceeds to step Sll, where 16 pixels of the next block are taken out from the original image storage device 22, and the process returns to step S2, where the above-described process is repeated. When all blocks have been processed, the process advances to step S12.

In this step Sl2, the pixel average value La, the difference value Ld
, the total code amount of the block identifier is obtained from each storage section in the RAM 24, and the code aS that can be assigned to the bit brain information φ2 is calculated by subtracting the total code amount from the upper limit of the code amount (step S12). .

In the block of mode B, all pit brain information φ2 becomes "0", but this can be determined from the block identifier, so there is no need to encode φ2. Mode C blocks require a maximum of 18 bits for bit plane information φ2 for each block. When dividing a block into mode B and mode C, blocks where D≦T2J holds are coded as mode B1, blocks where mode C does not hold are coded as mode C, and the D storage unit 24f
If the frequency of DI obtained from is 01, then S〉16Σn
1 (i is calculated from the larger one) and determines the value of T2 as T 2 "" D t - t (step S13). By determining the value of T2 in this manner, it is possible to obtain the best image quality while maintaining the preset code amount.

Next, the identifier for the first block is retrieved from the block identifier storage section 24a (step s14), and it is determined whether the mode is A or not based on this identifier. If it is determined that the mode is mode A, the process proceeds to step 521, and if it is determined that the mode is other than mode A, the process proceeds to step SIG. In this step Sl8, 16 pixels of the corresponding block are stored in the original image storage device 2.
2, the maximum 1ii MAXL and the minimum value M I NL in the block are determined (step S17), and it is determined whether MAXL-MINL≦T2 holds (step 818).

When the condition of step S18 is satisfied, step S18 is satisfied.
If the conditions are not met, it is determined that mode C is selected and the process proceeds to step Sl9.

In step Sl9, X+1 ≧ (MAXL+3M summer NL)/4,
and x. 1<C3MAXL+MINL)/4. (φ2) 1 for l is set to "0°, and (φ2),, for other pixels is set to 1 bit "1". Then, in order to change the block identifier to 1 bit "11" for mode C, the corresponding The contents of the block identifier storage unit 24a corresponding to the block are rewritten. Then, the bit brain information φ2 is encoded using the φ2 encoding table 23e and stored in the φ2 storage unit 24e (step S20).
.

After that, it is checked whether the processing from step 815 onwards for all blocks has been completed (step S21).
. If this step S21 is not established, the block identifier for the next block is stored in the block identifier storage unit 24.
Take out from a (step S22) and return to step S15 to repeat the above operation.

Then, when processing for all blocks is finished, the RAM
The block identifier stored in 24, the pixel average value La,
The difference value Ld and bit brain information φl, φ2 are transferred to the code word storage device 25.

This completes the block encoding process for one screen.

Then, on the decoding device side, a decoding process is performed based on the code word stored in the code word storage device 25, and the original image is decoded and displayed on the display device.

[Effects of the Invention] As described in detail above, according to the present invention, when an image signal is divided into blocks and encoded, the number of occurrences of the difference between the maximum value and the minimum value of pixels in the block is stored, and the number of occurrences of the difference between the maximum value and the minimum value of pixels in the block is stored, and the number of occurrences of the difference between the maximum value and the minimum value of pixels in the block is In addition to dividing into the optimal block mode, a block identifier corresponding to that mode is attached, and the average value of the pixels in the block, the difference chain,
Since the representative level is encoded, the amount of code can be kept constant and the processing can be simplified. In addition, since the amount of code can be made constant through simple processing as described above, it is not only possible to transmit or record television images or facsimile images, but also to electronically record image information of photographs taken on recording media such as floppy disks. It can be used in electronic still cameras and the like.

It is a flowchart which shows the processing operation of each mode A-C in a figure.

2l...CPU, 22...Image processing device, 23...
- Encoding table, 24...RAM, 25...code word storage device.

Claims (1)

[Claims] A dividing means for dividing an original image signal into a plurality of blocks; a block information detecting means for detecting the maximum value, minimum value, average value, difference value, and representative level of pixels in each block divided by the means; and this means. a counting means for counting the number of occurrences of the difference between the maximum value and the minimum value of the pixels in the block detected by the counting means; and a mode for determining the mode of encoding processing for each block based on the value counted by the counting means. determining means; means for adding a block identifier indicating the encoding mode of the block according to the determination result by the determining means; 1. A block encoding method comprising: encoding means for encoding.