JPH048086A - Coding system - Google Patents

Abstract

PURPOSE:To compress a data of a picture with high quality with a small hardware scale by providing a subscanning direction inverting input means which input picture information in the coding system applying thinning of the picture and inverting the scanning direction for each of L lines so as to apply simple variable density sampling. CONSTITUTION:An output of a selector 107, that is, a video signal whose scanning direction is reversed for each of L lines is inputted to a D latch 114. Then in the case of a picture block in which mode information is logical 1, that is, lots of high frequency components are included, and the normal clock is latched. When mode information is logical 0, an input picture signal is latched by a clock whose frequency is 1/4. A picture signal outputted to an output terminal 114 is formatted to an area of mode information and picture information and the result is sent. Then sent information is demodulated by a decoder. Thus, simple variable density sampling is implemented to apply data compression of a picture with high quality and small hardware scale.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to an encoding method, particularly to an encoding method for compressing data of an image signal such as a television signal that is composed of scanning lines. be.

[Conventional technology]

Figure 5 is a diagram explaining the basic concept of a conventional TAT (described later), Figure 6 is a diagram showing a data transmission pattern in a conventional two-dimensional TAT (described later), and Figure 7 is a diagram illustrating a data transmission pattern in a conventional two-dimensional TAT (described later). FIG. 8 is a block diagram of the transmitting side of a transmission system using dimensional TAT. FIG. 8 is a block diagram of the receiving side of a conventional two-dimensional TAT transmission system.

Conventionally, when transmitting information, the theme has always been how to reduce the amount of information to be transmitted so that the original information can be faithfully reproduced, and a wide variety of transmission methods have been proposed for this purpose.

For the above-mentioned theme, there is an adaptive town density sampler method in which the sampling density, that is, the information density to be transmitted, is changed as appropriate. An example of this method is the time axis conversion band compression method (Time Axis Tra), which has already been announced.
nsform (hereinafter referred to as TAT) will be explained using FIG.

FIG. 5 is a diagram for explaining the basic concept of TAT, and is a diagram for explaining information transmission by variable density sampling in the conventional example. In Figure 5, the original signal is divided into a predetermined period (predetermined amount of information) @ as shown by the dotted line, and it is determined whether the information contained in each divided group is coarse or dense. . Then, for the group judged to be dense, all the data obtained by sampling the original signal is transmitted as transmission data, and for the block judged to be coarse, only a part of all the data is used as transmission data.
Other data shall not be transmitted as thinned data.

The above concept reduces the number of data transmitted per unit time, making it possible to compress the bandwidth of the transmitted signal. The data transmitted in this manner is used on the receiving side to form data corresponding to the thinned-out data. That is, interpolated data that approximates the thinned-out data is calculated using the transmitted data. Since this interpolated data corresponds to the part where the information power i is coarse, it becomes data that is extremely similar to the thinned data. Therefore, compared to the case where all data is transmitted, the fidelity of the restored signal to the original signal hardly changes, and the transmission band can be significantly compressed. In other words, it is possible to reduce the amount of information to be transmitted.

On the other hand, the decision whether to transmit all the sample link data or a part of the data is made by checking the details of the original number 3, and this decision information is also based on the transmission mote information. simultaneously communicate in some form.

Now, let us consider applying the above-mentioned concept to the transmission of image information. Since image information has a two-dimensional spread and is correlated both horizontally and vertically, more effective transmission is possible if not only the horizontal sampling interval but also the vertical sampling interval can be varied. becomes.

This concept is hereinafter referred to as two-dimensional TAT, and stiffness will be briefly explained below. The concept of this two-dimensional TAT has already been disclosed in Japanese Patent Application No. 148112/1983 filed by the present applicant.

FIG. 6 is a diagram showing a data transmission pattern in the two-dimensional TAT. In two-dimensional TAT, one screen is divided into pixel blocks each consisting of m×n pixels, and the density of transmitted data is changed for each pixel block. In FIG. 6, it is assumed that the pixel block has 4×4 pixels, and two types of transmission modes are shown.

In the figure, O indicates a transmission pixel, and × indicates a thinning pixel. E
As shown in the figure, C shows a pattern in which all pixel data is transmitted, and C shows a pattern in which only a part of all pixel data is transmitted. Hereinafter, transmission modes using these transmission patterns will be referred to as E mode and C mode, respectively. As is clear from Figure 6, C mode is different from E mode! It can be seen that transmission is performed at an information density of /4.

Regarding the thinned out pixels of the pixel block transmitted in the C mode, the receiving side forms interpolated pixel data using adjacent pixel data among the transmitted pixel data to restore the original screen.

Next, transmission by two-dimensional TAT as described above will be explained using FIG. FIG. 7 is a diagram showing the configuration of the transmitting side of a transmission system using two-dimensional TAT. This case will be explained using a digital transmission system as an example.

In FIG. 7 of the drawing, the manually inputted bidet No. 4'3 is sampled for all pixels by an analog digital converter (A/D) 1 to generate all pixel data. When this all pixel data is supplied to the decimation circuit 2, after performing band limitation, decimation corresponding to the C-mote pattern shown in FIG. 6 is performed.
Obtain C-mote data. This C mote data is supplied to the interpolation circuit 3, and interpolated pixel data corresponding to the thinned out pixels is calculated.

This interpolated pixel data is supplied to the mote determination circuit 4 together with all the pixel data output from the A/D 1, and it is determined whether each pixel block is to be transmitted in C mode or E mode. The mote discrimination circuit 4 calculates the difference between the pixel data output from the A/D 1 and the interpolated pixel data, calculates the sum of this difference (hereinafter referred to as block distortion) for each pixel block, and calculates this for one frame. Store one word in memory.

Then, until the next field's take is input, the block distortion distribution of all pixel blocks is determined. Here, the ratio between the number of pixel blocks transmitted by C mode and the number of pixel blocks transmitted by C mode must always be constant. For example, pixel blocks transmitted by C mode are 2/3 of the total, and pixel blocks transmitted by C mode are 1/3 of the total.
If it is set to 3, the total number of data to be transmitted (compression rate) will be (2/3xl/4+1/3xl=) 1/2.

Therefore, based on the block distortion distribution of all pixel blocks, a distortion threshold value for determining the block distortion at which the C moat and C moat should be distributed is determined.

Then, at the timing when the next field's video signal is input, the stored block distortions are sequentially read out and compared with the distortion threshold to determine the transmission mode. If the read block distortion matches the distortion threshold, the transmission mode is set so that the ratio of the pixel blocks transmitted in C mode and the pixel blocks transmitted in C mode matches at a predetermined ratio as described above. It is determined.

The mote discrimination signal obtained in the child series is supplied to the switch 7, and the pixel take is alternatively read out from the buffer 5 for E mote take and the buffer 6 for C mote take. The output data of this switch 7 is sent out to the transmission path as transmission data.

Further, the mote discrimination signal is also outputted to the transmission line as mote information via the buffer 9.

Next, the operation on the receiving side will be explained using FIG.

FIG. 8 shows the configuration of the receiving side of the two-dimensional TAT transmission system. The digital hiteo signal input via the transmission path and subjected to the above-described processing is supplied to the C-mode interpolation circuit 11, where interpolated data corresponding to the thinned-out pixel data in the C-mode is calculated.

On the other hand, the transmitted mode information controls the switch 12,
When indicating mode information or C mode, connect to the E side, and when indicating C mode, connect to the C side.

As a result, all pixel takes including C-mode data, C-mode data, and interpolated pixel data are stored in the frame memory 13. All pixel takes are read out from the frame memory 13 in an order based on a television signal, for example, and outputted via a digital-to-analog converter (D/A) 14.

As explained above, in the two-dimensional TAT transmission system, image information can be transmitted extremely effectively.

(Problems to be Solved by the Invention) As explained above, in the conventional example, the transmission bit rate is set to a constant value by calculating the frequency of the sample pattern for each frame or field of the image. This requires a memory and a counter for counting the frequency of sample patterns, resulting in a problem that the size of the heart increases. Another problem is that it requires not only the heart scale but also a complex algorithm that completes the processing of stiffness within one frame period.

This invention was made to solve the above problems, and by reversing the scanning direction every L line and performing simple variable density sunblinking, it is possible to create high-quality images on a small heart scale. The object of the present invention is to obtain an encoding device that can perform data compression.

(Means for Solving the Problems) For this reason, the present invention provides a pixel thinning means for thinning out pixels in the sub-scanning direction in N cycles and M cycles, and a high definition information determination that extracts and determines high definition information of an image. and means;
In the encoding method in which the high-definition information determining means switches between the two pixel thinning means to thin out the image, the encoding method is performed while changing the direction every L lines in the sub-scanning direction and the opposite direction. The above object is achieved by an encoding system including a sub-scanning direction reversal input means for inputting image information.

(Function) The encoding method according to the present invention includes a pixel thinning means for thinning out pixels in the sub-scanning direction in N and M periods, a high-definition information determining means for extracting and determining high-definition information of the image, Based on this determination, the two pixel interpolation means are switched, and the sub-scanning direction reversal input means inputs image information while changing the direction every L lines between the sub-scanning direction and the opposite direction.

〔Example〕

Hereinafter, one embodiment of the present invention will be described based on the drawings.

Figure 1 is a diagram showing the configuration of an encoding system according to an embodiment of the present invention, Figure 2 is a diagram showing the format of an encoded image according to this embodiment, and Figure 3 is an image compressed using this embodiment. FIG. 4 is a diagram showing the compression sampling mode of this embodiment.

In FIG. 1, A is a pixel thinning means, B is a high-definition information determining means, C is a sub-scanning direction reversal input means, and Y is an encoding system 21o1 of this embodiment, which is an A/D converter (not shown). An input terminal for inputting the digitized TVi enlarged image signal, 102 is a timing generator, and 103 is controlled by the timing generator 102, and is turned on for each line.

The OFF switch 104 is also the timing generation a
105 is a random access memory controlled by the address generated by the uptown counter; 106 is a 1-line delay line, 10
7 is a selector (sub-scanning direction inversion input means) controlled by the timing generator 102, 108 is a horizontal low-pass filter, 109 is a subtracter, 110 is a switch for thinning out pixels (pixel thinning means), and 111 is a mote. Comparator that generates information (high-definition information judgment means),
112 is a counter that is reset for each line; 11
3 is a switch controlled by a counter flag, 1
14 is a switch for switching sampling mode; 11
5 is a selector that generates a mode switching signal, 116 is a 4-frequency divider counter, 117 is an output terminal that outputs a compressed image signal, and 118 is an output terminal that outputs Sunblink mote information.

Next, the operation of this embodiment will be explained using FIGS. 1 to 4.

First, in FIG. 1 of the drawing, the input terminal 101 has an A/D
A converter (not shown) converts a digital video signal such as a TV signal. Video signal is RAM1
05 and IH delay fi106. RAM105
is controlled by the address generated by the uptown counter 104 and the R/W (sword write) control signal from the timing generator 102. When the uptown counter is currently counting up, the control signal from the timing generator that writes the video signal in order from address O is generating a signal synchronized with the image whose level is inverted for each IH. When writing “l”
” on the next line, the control signal becomes “0”, and at this time, the counter 104 becomes a town count, and the RAM 105 becomes a read mode (read mode). Next, the signal from the RAM 105 and the IH delay line 106(7) signal is input to selector 107. Selector 107 is R
It is controlled by the same control signal as the read/write control signal of AM105. It is connected to select either the selector 107 or the RAM 105 side when the RAM 105 is read.

When the RAM 105 is in write mode, the selector 107 selects the IH delay line, so the pressure signal of the selector 107 can output a TV signal with the scanning direction reversed for each L line (in this case, 1 line). The output of the selector 107 is input to a low-pass filter 108, and its output is input to a subtracter 109.The other input of the subtracter 109 is input with a video signal that has not passed through the low-pass filter, and the difference between the two is input. The difference signal from the subtractor B109 is a signal representing the high-frequency components of the image, that is, the details.The output signal of the subtractor 109 is obtained by dividing the image sampling clock generated from the timing generator 102 into a frequency divider. 1/4 by switch 110 controlled by a signal divided by 4 by 116
The data are thinned out and manually input to the comparator 111.

The comparator 111 is reset every line, and the comparator 111 is reset every line, and the comparator 111 is reset for each line.
Counting flags from 11 onwards. counter 112
is calculated up to 1/3 of the pixels in one line,
When it becomes 1/3 or more, a flag of "1" is set. By doing this, even if there is a lot of detailed information about the picture in the first half of the line, only 1/3 of the pixels in one line will be in the high-skin moat (moat "l"), and the rest will be Mort
0" and can keep the transmission bit rate constant. The selector 113 is controlled by the flag of the counter 112, and when the flag is "1", it selects 0, and when the flag is "1", it selects the output of the comparator 111. It is output as mode information to the output terminal 118.The selector 115 receives the pixel sunblink clock and the clock whose frequency is divided by 4 using the frequency divider 116, and when the mode information is "1", the normal clock 1/4 if mode information is “0”
The divided clock is selected and output to control the switch 114.

The D latch 114 receives the output of the selector 107, that is,
A video signal with the scanning direction reversed is manually input for each L line (in this case, 1 line), and if the mote information is "1", that is, an image block (4 pixel unit) that contains many high-frequency components, it is processed using the normal clock. Latch, mote information is “0”
In this case, that is, in the case of an image block that does not include high-frequency components, the human-powered image signal is latched at 1/4 of the clock. The image signal outputted to the output terminal 114 is formatted into a mode information and image information area according to a format (not shown) as shown in FIG. 2, and then transmitted. The transmitted information is demodulated by the decoder shown in FIG. In FIG. 3, 302 and 304 are selectors, and 301
is an input terminal for image information, 312 is an input terminal for mote information, 303 is an interpolator for when there is no high frequency component of the image, 305
is an output terminal, which is separated into image information and mote information by a data divider (not shown), and is input to input terminals 301 and 312, respectively.

In the case of mode 1, it passes through each of the selectors 302 and 304 and is output as it is to the output terminal 305, and in the case of mode 0, the interpolator side is selected by the selector 302, interpolated by the interpolator 303, and the selector 304
is output through.

Then, the up-counter 1 shown in FIG.
04. RAM105. IH delay line 106 and selector 10
In the same way as in step 7, the scanning direction is also returned to the original, and the original image is decoded.

FIG. 4 shows the mode and image sampling pattern. In this example, the case where the image is compressed to 1/2 is described, so the state of mote "1" is 1/3 of the whole, and the state of mote "0" is 2/3 of the whole. The number is specified. That is, when the total number of pixels is 4, the ratio is 4 pixels for mote "1" and 1 pixel for mote "0". For this reason, k/3
x 4 + 2 k / 3 x 1 = 2 k
...... (1) and the target is 1/2 of all pixels.
I understand that the evening hours will be reduced.

〔Effect of the invention〕

As explained above, according to the present invention, by reversing the scanning direction every L line and performing simple variable density sampling, it is possible to perform high-quality image theta compression on a small heart scale. .

[Brief explanation of the drawing]

Fig. 1 is a diagram showing the configuration of an encoding system that is an embodiment of the present invention, Fig. 2 is a diagram showing the format of an image after encoding in this embodiment, and Fig. 3 is a diagram showing the format of an image compressed in this embodiment. FIG. 4 is a diagram showing the compression sunblink mode of this embodiment. FIG. 5 is a diagram explaining the basic concept of TAT in the conventional example. FIG. 6 is a diagram showing the two-dimensional TAT in the conventional example.
FIG. 7 is a configuration diagram of the transmitting side of a conventional two-dimensional TAT transmission system.
FIG. 8 is a block diagram of the receiving side of a conventional two-dimensional TAT transmission system. A: Pixel thinning means B: High-definition information determining means C: Sub-scanning direction reversal input means Y:
- Encoding method % type % Code indicates the same or equivalent part Canon Co., Ltd.

Claims (1)

[Claims] It has a pixel thinning means for thinning out pixels in the sub-scanning direction in N periods and M periods, and a high definition information determining means for extracting and determining high definition information of an image, and the high definition information determining means thins out the pixels in the two pixels. In an encoding method that thins out images by switching the thinning means, L is used in the sub-scanning direction and the opposite direction.
1. An encoding system comprising sub-scanning direction reversal input means for inputting image information to said encoding system while changing the direction for every line.