JPS62260492A - Picture signal band compression system - Google Patents

Abstract

PURPOSE:To execute a band compression with high compressibility to the input digital picture signal of an animation without the deterioration of reproduced picture quality by dividing the input digital picture signal into low-frequency component and high-frequency component and executing an inter-frame encoding. CONSTITUTION:When an input digital picture signal is set as a subsample circuit by a subsample circuit 1, the low-frequency component of the input digital picture signal is obtained and the low-frequency component is encoded by the first inter-frame encoding circuit 3. Therefore high compressibility is obtained by the inter-frame encoding of the signal which has a few samples. By executing a linear interpolation to the output of the subsample circuit 1 by means of an interpolation circuit 2, the number of the samples which is the same number of the input digital picture signal is obtained and the high-frequency component of the input digital picture signal is obtained when the interpolation output signal is substracted from the input digital picture signal by a subtracter. The high-frequency component is encoded by the second inter-frame encoding circuit 4. Thus even in the case of the animation, the high compressibility with the aid of the inter-frame encoding is obtained.

Description

[Detailed Description of the Invention] [Summary] An input digital image signal is divided into low frequency components and high frequency components, and interframe coding is performed on each of them.
It is also possible to perform band compression on input digital image signals of moving images at a high compression rate without deteriorating the reproduced image quality.

[Industrial application field]

The present invention relates to an image signal band compression method that can band-compress an input digital image signal at a high compression rate with a relatively simple configuration.

The band of the image signal is extremely wide compared to the band of the audio signal. Therefore, in order to reduce transmission costs, it has been adopted to compress the band for transmission. In that case, the image signal is divided between adjacent pixels in the same frame.
Alternatively, since there is a large correlation between pixels in the previous and subsequent frames, the band of the image signal can be compressed by utilizing such statistical properties. In addition, by taking advantage of the fact that humans have low visual sensitivity to high-frequency components and color components of image signals, making it difficult to recognize distortion in areas with large gradation changes, even if distortion occurs in the reproduced image, it is not a visual disturbance. Another possible method is to compress the bandwidth by reducing information that would otherwise not be available.

As such a band compression method, predictive coding (P r
editive coding) and orthogonal transform coding (
Orthogonal transform cod
ing). The former predictive encoding method quantizes and encodes the difference (prediction error) between the current pixel value of the input image signal and the value predicted from the past pixel value (predicted value), and is called differential encoding. Also called. The latter orthogonal transform encoding method divides the screen into blocks of a certain size each consisting of multiple pixels, performs orthogonal transform on each block, and quantizes and encodes each component of the transform output. It is something.

In addition, predictive coding and orthogonal transform coding are divided into intraframe coding, which processes using pixels within the same frame, and interframe coding, which processes using pixels between adjacent frames. be able to. Also known is a composite differential encoding method that combines an intraframe encoding method and an interframe encoding method.

Generally, predictive coding methods are often used. In this predictive coding method, the prediction error becomes an extremely small value compared to the original pixel value, and the prediction error amplitude distribution concentrates at a small amplitude level near zero. Utilizing the statistical properties of such prediction errors, fine-grained calculations can be made for small amplitude levels.
Nonlinear quantization means that performs quantization with coarse settings for large amplitude levels can reduce the number of quantization levels. On the other hand, linear quantization means that performs quantization in fixed quantization level steps has a simpler configuration than nonlinear quantization means.

In addition, the input analog image signal is converted into a digital signal and subjected to band compression processing, and even when transmitting moving images, it is desired to obtain a high compression rate with a simple configuration.

[Conventional technology]

As described above, various methods have already been proposed as band compression methods for image signals. For still images or similar images with little movement, the correlation between frames is large, so it is possible to significantly compress the bandwidth by interframe coding. Furthermore, in the case of moving images, since the correlation between frames is small, intraframe coding has a higher compression rate than interframe coding, so intraframe coding is used. As described above, a method is also known in which it is determined whether or not there is movement based on the correlation value between frames, and the method is switched between interframe encoding and intraframe encoding.

Furthermore, in the case of a moving image, the background and the like are similar to those of a still image, so a method is also known in which image signals of only moving parts are transmitted. That is, after transmitting an image signal for one screen, the inter-frame difference values of only moving areas are quantized and encoded,
An address code indicating the location of the area is added and sent. On the receiving side, the received image signal for one screen is written in a screen memory with a storage capacity for one screen, and the contents of the screen memory are played back and displayed, and when the image signal of the moving area to which the address code is added is received. , the contents of the area of the screen memory indicated by the address code are rewritten with the received image signal to reproduce and display a moving image, and areas with no movement are repeatedly reproduced and displayed.

[Problem that the invention seeks to solve]

When performing band compression on a still image or an image that contains some moving areas, it is better to perform band compression than intra-frame encoding.
A high compression rate can be obtained by performing interframe encoding. However, when performing band compression on images that include moving areas, such as videos, intra-frame coding is more capable of band compression than inter-frame coding; The compression rate is lower than when band compression is performed by interframe coding.

Furthermore, in a method of transmitting an image signal of only a moving area, since the band of a relatively narrow moving area is compressed and transmitted, it is possible to lower the transmission bit rate. However, this method has the disadvantage that the function for identifying moving areas and stationary areas is complicated, and the process of specifying and transmitting the area is complicated.

An object of the present invention is to provide an image signal band compression method that can obtain a high compression rate with a relatively simple configuration.

[Means for solving problems]

The image signal band compression method of the present invention performs interframe encoding on low frequency components and high frequency components of an input digital image signal, and will be explained with reference to FIG. A sub-sampling circuit 1 that sub-samples an input digital image signal, and an interpolation that linearly interpolates the output signal sub-sampled by this sub-sampling circuit 1 every other pixel so that it has the same number of samples as the input digital image signal. circuit 2 and first and second interframe coding circuits 3.4
and a subtracter 5, the output signal of the sub-sampling circuit 1 is encoded by the first inter-frame encoding circuit 3, thereby outputting an inter-frame encoded signal of the low frequency component of the input digital image signal. The input digital image signal and the output signal of the interpolation circuit 2 are added to the subtracter 5, and the output signal of the difference is encoded by the second interframe encoding circuit 4, thereby converting the high frequency of the input digital image signal. The interframe encoded signal of the component is output, and the interframe encoded signal of the low frequency component and high frequency component of the input digital image signal is sent to the receiving side.

[Effect]

When the input digital image signal is subsampled by the subsampling circuit 1, a low frequency component of the input digital image signal is obtained. This low frequency component is encoded by the first interframe encoding circuit 3. Therefore, a high compression rate can be obtained by interframe encoding a signal with a small number of samples.

Furthermore, the output signal of the sub-sampling circuit 1 is linearly interpolated by the interpolation circuit 2 to obtain the same number of samples as the input digital image signal, and the interpolation output signal is subtracted from the input digital image signal by the subtracter 5. High frequency components are obtained. This high frequency component is encoded by the second interframe encoding circuit 4. Therefore, interframe encoded signals can be obtained for the low frequency components and high frequency components of the input digital image signal, and even in the case of moving images,
A high compression rate can be obtained by interframe coding.

〔Example〕

Embodiments of the present invention will be described in detail below with reference to the drawings.

FIG. 2 is a block diagram of the main parts on the transmitting side of the embodiment of the present invention, in which 11 is an A/D converter that converts the analog image signal a into, for example, an 8-bit digital image signal, and 12 is a sub-sampling circuit. 13 is an interpolation circuit, 14 to 16 are subtracters, 17.18 is an adder, 19.21 is a quantizer, 20
．． 22 is a frame buffer. Subsample circuit 1
2 corresponds to the sub-sample circuit 1 in FIG. 1, the interpolation circuit 13 corresponds to the interpolation circuit 2 in FIG. 1, and the subtracter 14. Adder 17. The child converter 19 and the frame buffer 20 correspond to the first interframe encoding circuit 3 in FIG. 1, and the subtracter 16 . The adder 18, quantizer 21, and frame buffer 22 correspond to the second interframe encoding circuit 4 in FIG.

The analog image signal a obtained from a television camera etc. is A
The /D converter 11 converts the signal into an 8-bit digital image signal with, for example, 256 gradations, and applies it to the sub-sampling circuit 12 and the subtracter 15.

The sub-sampling circuit 12 extracts, for example, every other sample from the sample value sequence of the digital image signal S, and its output signal C represents the low frequency component of the digital image signal S. . The output signal C is applied to an interpolator 13 and a subtracter 14.

The interpolation circuit 13 performs linear interpolation processing and applies its output signal to the subtracter 15. Therefore, the interpolation output signal is subtracted from the digital image signal by the subtracter 15, and the output signal d has high frequency components only for steep changes in the digital image signal.

In the subtracter 14 of the interframe encoding circuit of the writing 1,
The read signal e (prediction value) of the frame buffer 20 is subtracted from the output signal C of the sub-sample circuit 12, and the output signal f (prediction error) is added to the quantizer 19. The quantizer 19 is a Laplace quantizer that uses the statistical properties of prediction errors, and quantizes the prediction errors, and its output signal g is an interframe encoded signal of the low frequency component of the digital image signal. becomes. Also, the output signal g of the quantizer 19 is
The adder 17 adds the signal e to the subsequent signal e from the frame buffer 20, and writes it into the frame buffer 20 as a predicted value for the next frame.

In this case, since the input digital image signal is a low frequency component and the number of samples is small, it will be interframe encoded at a high compression rate.

In the subtracter 16 of the interframe encoding circuit of the writing brush 2,
The subsequent signal h (prediction value) of the frame buffer 22 is subtracted from the output signal d of the subtracter 15, and the output signal i (prediction error) is added to the quantizer 21. This quantizer 2
Similarly to the quantizer 19, the quantizer 1 is also a Laplace quantizer, and the quantization step is set to suit the nature of the high frequency component of the digital image signal. The quantizer 21 quantizes the prediction error, and its output signal j becomes an interframe encoded signal of the high frequency component of the digital image signal. Then, the adder 18 adds the output signal j of the quantizer 21 and the subsequent signal from the frame buffer 22, and writes the sum into the frame buffer 22 as a predicted value for the next frame. In this case, the output signal d of the subtracter 15 as a high frequency component of the digital image signal S is
Since the amplitude level is small, interframe encoding is performed at a high compression rate.

The output signals g and j of the quantizers 19 and 21 are sent as interframe encoded signals to the receiving side in a time division multiplex format, orthogonal modulation format, or the like.

FIG. 3 is a block diagram of the main parts of the receiving side of the embodiment of the present invention, in which 30 is a receiving section, 31, 32, 33 is an adder, 34.
35 is a frame buffer, and 36 is an interpolation circuit.

A signal k from the transmitting side is received by the receiving unit 30, and interframe encoded signals g, j of low frequency components and high frequency components from the transmitting side are received.
Encoded signal corresponding to 1. Separate into m. Those encoded signals 1. m are added to adders 31 and 32, respectively.

The received and separated low frequency component encoded signal l and the signal (predicted value) read from the frame buffer 34 are sent to the adder 3.
It is added by 1 and written to the frame buffer 34 as a predicted value for the next frame, and is also added to the interpolation circuit 36. In this interpolation circuit 36, since the low frequency component encoded signal l is subsampled and the number of samples is reduced as described above, linear interpolation processing is performed.
It is added to the adder 33 as a low frequency component signal n having the same number of samples as the digital image signal on the transmitting side.

Further, the received and separated high frequency component encoded signal m and the signal (predicted value) read from the frame buffer 35 are added by the adder 32, and the high frequency component encoded signal m is added to the adder 3 as the high frequency component signal p.
3 and written into the frame buffer 35 as the next prediction signal.

The low frequency component signal n and the high frequency component signal p are added by an adder 33, and a digital image signal q is output.

This digital image signal q corresponds to the digital image signal S on the transmitting side, and is converted into an analog image signal by being applied to a display device (not shown) and reproduced and displayed on the display section.

〔Effect of the invention〕

As explained above, the present invention subsamples a human-powered digital image signal using the subsampling circuit 1 to obtain a low frequency component, and performs interframe coding on this low frequency component, thereby reducing the number of samples. Therefore, it is possible to perform band compression with a high compression rate.

Furthermore, the interpolation circuit 2 performs linear interpolation processing on the subsampled input digital image signal, subtracts it from the input digital image signal to obtain a high frequency component, and performs interframe coding on this high frequency component. If only the high-frequency components are used, the reproduced image quality of moving parts deteriorates, but since this high-frequency component is also compressed and encoded by interframe encoding and transmitted, deterioration of the reproduced image quality can be prevented. Also, since the difference between the input digital image signal and the linearly interpolated signal is used as a high frequency component signal,
The high frequency component signal has a small amplitude level, and interframe encoding can be performed at a high compression ratio. Therefore,
This has the advantage that there is little deterioration in reproduced image quality and that a high compression ratio of about 4:1 can be achieved with a relatively simple configuration.

Since a high compression rate can be obtained with such a relatively simple configuration, it can be applied to band compression of image signals for videophone calls, video conferences, and the like.

[Brief explanation of drawings]

FIG. 1 is a block diagram of the principle of the present invention, FIG. 2 is a block diagram of essential parts on the transmitting side of an embodiment of the invention, and FIG. 3 is a block diagram of essential parts on the receiving side of an embodiment of the invention. 1 is a sub-sampling circuit, 2 is an interpolation circuit, 3.4 is the first and second interframe coding circuit, 5 is a subtracter, 11 is A
/D converter, 12 is a sub-sampling circuit, 13 is an interpolation circuit, 14, 15.16 is a subtracter, 17.18 is an adder, 1
9.21 is a quantizer, and 20.22 is a frame buffer.

Claims (1)

[Claims] A sub-sampling circuit (1) that sub-samples an input digital image signal, an interpolation circuit (2) that linearly interpolates the output signal of the sub-sampling circuit (1), and first and second frames. inter-coding circuits (3) and (4), applying the output signal of the sub-sample circuit (1) to the first inter-frame encoding circuit (3); (3) outputs an interframe encoded signal of the low frequency component of the input digital image signal, and outputs a signal of the difference between the input digital image signal and the output signal of the interpolation circuit (2) as a signal between the second frames. In addition to the encoding circuit (4), the second interframe encoding circuit (4) outputs an interframe encoded signal of the high frequency component of the input digital image signal, and outputs an interframe encoded signal of the high frequency component of the input digital image signal and the low frequency component of the input digital image signal. An image signal band compression method characterized by sending an interframe coded signal with high frequency components to the receiving side.