JPS6023543B2 - Signal transmission method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a signal transmission method using predictive encoding and code conversion, and in particular, the transmission code format is switched depending on the nature of the gradation of characters, photographs, etc.

Generally, when transmitting an image having gradation, band compression is achieved by predictive encoding and code conversion. However, the images to be transmitted include text, photographs, and composite forms thereof, and each form has the disadvantage that the gradation properties are not constant, resulting in a reduction in compression efficiency. Therefore, the present invention aims to further improve compression efficiency by preparing a plurality of types of code tables and selecting these code tables according to the nature of image gradation.

Examples will be described below. The figure is a block diagram showing one embodiment of the present invention.

Reference numeral 10 indicates a controller that controls the entire system, and reference numeral 11 indicates a memory for one scan of the image signal Sa obtained by converting one pixel into 4 bits. , 1 is calculated by a subtracter 13 serving as a prediction calculation unit.

Match circuits 14 to 16 determine whether the 5-bit prediction error signal Sb of the subtracter 13 is equal to 14, 19, and 113 respectively set. If it is equal to either
One of the corresponding match signals Sa to Sc is output. 17 to 19 are counters for counting the generation circuits of the coincidence signals Sa to Sc, respectively, and Sf to Sh are count values of the respective counters 17 to 19, which are input to the comparator 20.

The controller 10 repeats this for each scan in the image memory 11. When one scan is completed, the comparator 20
The count values Sf to Sh of the counters 17 to 19 are compared and classified into the following three states. A: Sf, Sg, Sh
>Q (however, Q is a constant value).

B: Sf>Sg, Sh>Sg and Sg<8 (however, 6
is a constant value). C: Others.

Prediction error for one scan of characters, figures, and photographs -15
The distribution of ~15 has a peak at 0 billion, but in the case of figures with excellent contrast, it appears almost evenly in the range of prediction error values of 14 to 113, and the comparison circuit 2 detects state A. do.

In the case of characters and figures that can be counted as binary, the prediction error value is extremely small at 19, and the comparison circuit 20 detects state B.

In the case of figures or photographs with poor contrast, the state is C. The comparator 20 outputs any of these states A to C as a comparison result Sj to the controller 10, and the controller 10 sends this comparison result Si as 2-bit information to the transmission memory 21, and also sends the comparison result Si to the transmission memory 21 as 2-bit information. - Select one of the table. Next, the controller 1 reads out each pixel from the image memory 11 for transmission, and the subtracter 1
3 outputs the prediction error signal Sb pixel by pixel to the encoding circuit 22.

The encoding circuit has the following three types of code tables A, B,
C are prepared, and one of them is in a selected state.
Co. Dotaple A code table B code table 0
Prediction error output code Prediction error output code Prediction error output code 11510010・01511511100011
0111410011014114111000110
1013101000131131110001111
2101010121121010010111000
000111111110011○100001010
1101100019100010091911000
0810001108181100117101100
707110○106101110606110101
5110000505101100411001040
4101011311010030310100021
10110202101111111010110○1
00〇〇〇〇-11111-10-11000-211
0111-20-211011-3110101-30
-3101010-4110011-40-41011
01-5110001-50-5110100-610
1111-60-6111010-7101101-7
0-7111011-81000111-81-811
1110-91000101-91-9111101-
101000011-101-10111100-11
1000001-111-11111111-1210
1011-121-121010011-131010
01-131-131110000-14100111
-141-1411100010-15100101-
151-151110001100 encoding circuit 22,
Code S according to the selected code table A, B, C
k is outputted to the transmission memory 21, and from the transmission memory 21 is outputted as a transmission signal G together with information on the previous comparison result Si.

The code table A is a variable length code, and its code length is approximately inversely proportional to the distribution of prediction error values for one scanning line in a figure with excellent contrast. By selecting case A, compression efficiency can be increased. The output code of code table 8 is a fixed length code of 200 million, and state B
In this case, it is possible to transmit the image with compression efficiency equivalent to that of a two-tone image.

The output code of code table C is a variable length code, and the code length is inversely proportional to the distribution of prediction error values for one scanning line in figures or photographs with poor contrast, so it is not always possible to achieve appropriate compression efficiency. Although it is not the one obtained, it is representative of the properties of other images.

As explained above, since the most efficient code table is applied to the nature of the input information, high compression efficiency can be obtained for a variety of input images. In the above embodiment, the output S of the transmission memory 21 is sent each time the encoding process is completed, but in the case of facsimile, two transmission memories 21 are prepared, and the transmission period and storage period are It is better to switch between each scanning line.

In addition, if it takes a relatively long time to write the image signal Sa to the image memory 1, the image memory 11
It is preferable to prepare two of them and to switch the write period and the write period for each scanning line. Furthermore, since the frequency of occurrence of prediction error value 0 in image signals is generally very high, run-length encoding is performed only for prediction error value 0, and the code conversion of the present invention is applied to the remaining prediction error values. may be done. In addition, as another method, the prediction error signal may be
Segment each pixel, send binary values "0" and "1" indicating whether or not the prediction error value consists of only 0, and perform the code conversion of the present invention only on the segment containing the prediction error value other than 0. Good too. Furthermore, in the embodiment described above, the selection of the code table is switched for each scanned image, but the same effect can be obtained by dividing one scanning line every 8 pixels into smaller blocks and switching for each block.

The same effect can be obtained by switching for each image of one screen. Furthermore, if the controller 10 is slow, it is preferable to obtain the comparison result Sj for a group of pixels and use the same code table for a large group of pixels including that group. Furthermore, in the above embodiment, the values of the matching circuits 14 to 16 are 1.
We considered +4, 19, and 113, but it would be better to take multiple matches, such as 14 and 15, 18 and 19, and 112 and 113, for frequency detection. As reliable. Since the present invention changes the code table depending on the input image, it is possible to obtain high transmission efficiency for a wide variety of input information, so it can be used in facsimiles, televisions, image storage devices, and the like.

[Brief explanation of drawings]

The figure is a block diagram of one embodiment of the present invention. 10... Controller, 11... Image memory, 12
...1 pixel memory, 13...subtractor, 14-16.
... Matching circuit, 17-19... Counter, 20...
Comparator, 21...Transmission memory, 22...Encoding circuit, Sa.
...Image signal, Sb...Prediction error signal, Si...Comparison result, Sk...Output code, Sg...Transmission signal.

Claims (1)

[Claims] 1. In a signal transmission system that transmits an image signal by code-converting a prediction error signal obtained by predictive coding of an image signal according to a code table, the code table may be a fixed-length code table or a variable-length code table. We offer multiple types of code tables, including code tables.
The frequency of occurrence of a plurality of specific prediction error signals in a group of pixels is compared, and code conversion of the prediction error signal in the group of pixels or the prediction error signal in a group of pixels including the group is selected based on the comparison result. A signal transmission method characterized by using a code table.