JPH0262175A - Picture encoding/transmission equipment - Google Patents

Abstract

PURPOSE:To prevent the disorder of a reproduced picture by separating picture signals into the both components of low frequencies and high frequencies by means of power distribution, and separately encoding and transmitting them. CONSTITUTION:The picture input signals are converted into frequency components by an orthogonal converting means 31, and a power distributing condition is calculated by a power calculating means 32. Based on this, a control means 34 determines the separating rate of the both low and high frequency components, and separates them by a separating means 33. When the generating information of the high frequency components is increased due to the characteristic of the picture, the separating boundary of the means is shifted to a low frequency side so that the information quantity of the low frequency components to a first encoding means 35 may be increased and the burden of a second encoding means 36 may be reduced. Further, when the generating information quantity of the low frequency components is increased, the boundary of the means 33 is shifted to the high frequency side, and the high frequency component information quantity to the second encoding means 36 is increased. Thus, the decrease of the encoding efficiency and the reproduced picture disorder can be prevented.

Description

[Detailed Description of the Invention] [Summary] The present invention relates to an image encoding and transmitting device that performs high-efficiency encoding of images.

The purpose is to adaptively respond to the characteristics of the image and prevent a decline in encoding efficiency, while at the same time preventing as much as possible the distortion of the reproduced image due to transmission errors, etc.

Orthogonal transformation means for orthogonally transforming an image signal into frequency components; power calculation means for calculating the power distribution of the frequency components transformed by the orthogonal transformation means; and frequency components transformed by the orthogonal transformation means into low frequency components and high frequency components. Separation means for separating, control means for changing the separation ratio of low frequency components and high frequency components in the separation means based on the power calculation result in the power calculation means 9. The low frequency components separated by the separation means are encoded with high coding efficiency. a first encoding means for encoding, and 1
a second encoding means for encoding the high frequency component separated by the separation means in a manner that allows fast attenuation of generated errors;
．． The output signal of the second encoding means is configured to be transmitted through two channels.

[Industrial application field]

The present invention relates to an image encoding and transmitting apparatus that performs high-efficiency encoding of images.

[Conventional technology]

As image encoding apparatuses that encode images with high efficiency, those using an interframe encoding method or those using an intraframe encoding method are known.

The interframe coding method has high coding efficiency and is commonly used in high-efficiency coding devices, but the reference screen on the receiving side is destroyed due to information discarded due to congestion in the transmission path, etc. If this happens, it will be difficult to play the subsequent screens. In such a case, the receiving side generally issues a request to switch to forced intraframe encoding mode to the transmitting side, and the encoding method is forcibly switched from interframe encoding to intraframe encoding on the transmitting side. A signal resulting from intraframe encoding of the screen is sent to the receiving side, the screen is reproduced, and then the mode is switched back to interframe encoding mode.

On the other hand, since the intraframe coding method can quickly attenuate the effects of transmission errors, it is easy to return to the original screen even if the screen is corrupted on the receiving side due to an error, etc.; however, the coding efficiency is limited between frames. It is inferior to the encoding method.

[Problem to be solved by the invention]

When a high-efficiency image signal encoding device transmits encoded image signals using a burst multiplex transmission network such as packet switching, transmission frames may be lost (packet discarded) due to transmission errors or buffer overflows. It is inevitable that it will occur. In this case, since a high-efficiency encoding device for one image signal generally performs high-efficiency encoding based on the difference between image frames, there is a problem in that all subsequent received signals become invalid and one screen collapses.

Furthermore, in order to restore this situation to normal,
A request to switch to forced intraframe coding mode is required from the receiving side to the sending side, and this forced intraframe coding has lower coding efficiency than interframe coding, so transmission takes time and reception The playback screen on the side temporarily freezes.

A problem has arisen in which serviceability deteriorates.

As a means to solve this problem, a two-channel encoding transmission system has been proposed in which one image signal is separated into low frequency components and high frequency components, each of which is encoded and transmitted separately.

FIG. 4 is a block diagram of an image encoding and transmitting apparatus using the two-channel encoding and transmitting method. In Figure 4, 21
is an orthogonal transformer 92 that orthogonally transforms the image signal into frequency components.
2 is a coefficient separator that separates the frequency component transformed by the orthogonal transformer 21 into a low frequency component and a high frequency component, and 23 is a complete coefficient separator that encodes the low frequency component separated by the coefficient separator 22 with high coding efficiency. An integral type interframe encoder 24 is a leaky integral type interframe encoder (or an intraframe encoding method) that encodes the high frequency components separated by the coefficient separator 22 using an encoding method in which generated errors are quickly attenuated. ), 25 has a channel CH with a low discard rate and a channel C1]2 with a high discard rate, and transmits the low frequency component encoded by the fully integral interframe encoder 23 on the channel CH with a low discard rate,
7! S! This is a multiplexing device that transmits high frequency components encoded by the integral type interframe encoder 24 through a channel with a high discard rate.

The operation of this image encoding and transmitting device will be explained below. An orthogonal transformer 21 converts the image signal into frequency components, and a coefficient separator 22 converts the image signal into low frequency components [L and high frequency components].
++ and are encoded separately. In this case, the low frequency components include information regarding the basic skeletal structure of the image.

On the other hand, the high frequency component element u contains information regarding fine parts of the image. Therefore, if the low frequency component "L" is lost, the reproduced image will be distorted, but on the other hand, the crystal frequency component "H" will be lost.
Even if only one image is lost, the image will only become unclear, and all five images will not be destroyed.

Therefore, the more important low frequency component! 5 by a fully precise interframe encoder 23 with high encoding efficiency.

In addition, the high-frequency component elements 11 are individually encoded using leakage integral type interframe encoding ``524,'' and multiplexed into the multiplexed bag;
5, the signals are sent to the receiving side via separate channels CH, , CH□, respectively.

Since the channel CH is designed to have a low discard rate even when the transmission path is congested, low frequency components that require more IR have a low probability of being lost even if information is discarded on the transmission path. Become.

In other words, even if the high-frequency component n transmitted through channel CH2, which has a higher discard rate, is lost due to information being discarded on the transmission path, the low-frequency component n is sent to the receiving example, so the image cannot be processed on the receiving side. can be prevented from being destroyed as much as possible.

In addition, since the high frequency component "11" is encoded using the intra-frame coding method, for example, where the attenuation of errors occurring at 2 is fast, even if the high frequency component "H" is lost in the transmission path, the high frequency component "H" becomes normal again after that. If "H" is sent, the playback screen can quickly return to the original screen, and successive returns are possible without the need for the receiving side to request the transmitting side to switch to forced intra-frame encoding mode.

On the other hand, in this proposed image coding and transmission device.

There are the following problems. That is, the coefficient separator 22
The separation ratio between the low frequency component L and the high frequency component H is fixed. Therefore, if the generation information of the high frequency component H increases due to the nature of the 9 images, this high frequency component This is because encoders that encode H generally have low encoding efficiency.

The overall encoding efficiency decreases. On the other hand, when the amount of generated information of low frequency components increases, the discard rate in the channel CH on the low discard rate side increases, and the probability that the reproduced image is corrupted increases.

Therefore, an object of the present invention is to adaptively respond to the characteristics of two images to prevent a decrease in encoding efficiency, and at the same time to prevent as much as possible the distortion of reproduced images due to transmission errors and the like.

[Means to solve the problem]

FIG. 1 is a principle block diagram according to the present invention.

In one embodiment of the image encoding and transmitting apparatus according to the present invention, orthogonal transformation means 3 orthogonally transforms one image signal into frequency components.
1. power calculation means 32 for calculating the power distribution of the frequency components transformed by the orthogonal transformation means 31; Orthogonal transformation means 31
T: Separator that separates the converted frequency component into a low frequency component and a high frequency component Vi33. Based on the power calculation result by the power calculation means 32, the control means 342 changes the separation ratio of the low frequency component "L" and the high frequency component sI in the separation means 33. First encoding means 3 encoding with high encoding efficiency
5. and a second encoding means 36 that encodes the high frequency component component ++ separated by the nine separation means 33 using a method (for example, leaky integral type interframe encoding or intraframe encoding) in which the generated errors are quickly attenuated. Equipped with 1. The output signal of the second encoding means 35 and 36 is configured to be transmitted through two channels.

Further, the image encoding and transmitting apparatus according to the present invention includes an orthogonal transform means 31 for orthogonally transforming one image signal into frequency components in the form of a block. power calculation means 3'2 for calculating the power distribution of the frequency components transformed by the orthogonal transformation means 31; Separation means 33 for separating the frequency components transformed by the orthogonal transformation means 31 into low frequency components and high frequency components o. Based on the power calculation result by the power calculation means 32, a control means 341 changes the separation ratio of low frequency components and high frequency components n by the separation means 33; A first encoding means 352 that encodes the high frequency component n separated by the separation unit 33 with a high efficiency of encoding. and has a channel CH with a low discard rate and a channel CH2 with a high discard rate, and transmits the low frequency component residue encoded by the first encoding means 35 through the channel CH with a low discard rate, and performs second encoding. It comprises a multiplex transmission means 37 for transmitting the high frequency component "H" encoded by the means 36 on a channel CH2 with a high discard rate.

[Operation] The image input signal is converted into frequency components by the orthogonal transform means 31, and the power distribution state of the frequency components is calculated by the power calculation means 32. Based on this calculation result, the control means 34 determines the separation ratio (boundary) between the low frequency component m and the high frequency component m by the separation means 33. A method for determining this boundary is 5.For example, the ratio of the power of the low frequency component H and the high frequency component H is calculated as
It is made to be constant according to the waste rate in step 2.

As a result, if the amount of generated information of high frequency components increases due to the nature of one screen, the amount of information of low frequency components encoded by the first encoding means 35 is increased and The separation boundary in the 9-separation means 33 is moved to the lower frequency side so as to reduce the burden on the 9-separation means 36.

On the other hand, when the amount of generated information of low frequency components increases due to the nature of the 9 screen, the amount of information input to the first encoding means 35 is reduced to suppress the increase in the information discard rate of the channel CH1. The separation boundary in the separation means 33 is moved to the higher frequency side to increase the amount of emotion of the high frequency component element n input to the second encoding means 36.

As a result, it is possible to adaptively deal with the characteristics of the image while preventing a decrease in encoding efficiency, and to prevent the playback screen from collapsing due to information being discarded at @1 as much as possible.

〔Example〕

The present invention will be described in detail below with reference to three drawings.

FIG. 2 is a block diagram showing an image encoding and transmitting apparatus as an embodiment of the present invention. In Fig. 2, l is a discrete number that performs two-dimensional m-losing cosine transform (DCT) on the input signal in block units (e.g., 8 x 8 pixels) to create blocks of frequency component transform coefficients as shown in Fig. 3. It is a cosine transformer. As is clear from FIG. 3, the low frequency components of the transform coefficients are distributed toward the upper left side of the block, and the high frequency components are distributed toward the lower right side of the block.

A block line buffer 2 stores the output signal of the discrete cosine transformer 1 for one block line. 3 is a power calculator, which calculates the power of each frequency component (that is, each coefficient) in FIG. 3, respectively. 5 is a subtracter that calculates the difference between the input means and the predicted value.

6 is a coefficient separator, and the 0 separation method for separating the block of transform coefficients transformed by the discrete cosine transformer l into a low frequency component fL and a high frequency component H is 2, for example, as shown in FIG. , boundary A.

is separated into a low frequency component t and a high frequency component n.

The separated low frequency component t is encoded by a fully integral type interframe predictive coding circuit, while the high frequency component u separated by one is coded by a leaky integral type interframe predictive coding circuit.

This fully integrated interframe predictive coding circuit.

Quantizer 7. Reverse ninetizer 9. Adder 11. Coefficient synthesis 31
4. Coefficient frame memory 15. Coefficient separator 6, +6. It is composed of a closed circuit including a subtracter 5 and the like.

- On the other hand, the leakage integral type interframe predictive coding circuit is.

Quantizer 8. Reverse doshi conversion device 10. Adder 12. Prediction coefficient multiplier 13. Coefficient synthesizer 14. Coefficient frame, 'Mo171
5. Coefficient #rI6.16. It consists of a closed circuit including MWWS2. The prediction coefficient multiplier 13 adds Pj#12
This is a circuit that multiplies the output signal of 1 by a prediction coefficient α (0 < α × 1). Therefore, the loop including this prediction coefficient multiplier 13 is a leaky integration type circuit.

The output signal of the fully integral type interframe predictive coding circuit that encodes the low frequency components (i.e., the output signal of the quantizer 7) is input to the encoder 17 and converted into a code format according to the configuration of the transmission path. encoded.

Further, the output signal of the leaky component type interframe predictive coding circuit that encodes the high frequency component s (i.e., the output signal of the quantizer 8) is input to the encoding 318 and encoded into a code format according to the transmission path configuration. be converted into

The output signals of these encoders 17, 18 are sent to a multiplexer. The multiplexer has two channels C for signal transmission.
111 and CH2, the channel CH1 is a channel with a low discard rate of information discarded when information is congested on the transmission path, while the channel CH2 is a channel with a high discard rate. Here, signals with low frequency components are connected to channel CI with a low discard rate to prevent them from being discarded as much as possible.5 On the other hand, signals with high frequency components ``)1'' are connected to channel C) with a high discard rate. When discarding information, it is discarded before low frequency components.

Reference numeral 4 denotes a controller, which changes the ratio of separation between the low frequency component + and the high frequency component 1H in the coefficient separator 6.16, based on the calculation result of the power meter W-device 3. For example, as shown in FIG. 3, the one-separation boundary Δ1 may be moved to A2 or A3.

In addition, the boundaries of coefficient synthesis in the coefficient synthesizer 14 are also controlled accordingly.

The operation of the example device will be described below.

The image signal is input to a discrete cosine transformer l, where a two-dimensional discrete cosine transform is performed on a block-by-block basis, and the signal is converted into frequency components. This result is shown once in block line units.
Dropped into a bronc line buffer. As a result, coefficient separation control is performed on a block line basis.

Next, the subtracter 5 calculates the difference between the coefficient read from the block line buffer 2 and the predicted value from the coefficient frame memory I5. Coefficient! The 91t unit 6 separates the output signal from the subtracter 5 into a low frequency component IL and a high frequency component "H" and outputs them.

On the other hand, in the power calculator 3, the block line buffer 2
Calculate the power of each frequency component from the output signal from. ;The controller 4 then calculates 2 based on the result of this power calculation.
For example, the boundaries of the yM wavenumber separation of the transform coefficient block in Fig. 3 are set so that the ratio between the low frequency component "L" and the high frequency component component I+ is a constant value determined by the ratio of the discard rates of channels CH and CH2. The boundary of separation in coefficient separator 6.16 and the boundary of synthesis in coefficient combiner 14 are controlled based on the determination result.

Examples of separation methods include 2. For example, as shown in Figure 3, when the amount of generated information of high frequency components of the input image becomes excessive, the separation boundary is moved to boundary A3 (jl'J) and a complete integral type The amount of information processed by the interframe predictive coding circuit is increased to improve coding efficiency, and if the amount of information generated in the low frequency components of the input image becomes excessive, it is moved to the boundary A2 and complete integration is performed. To reduce the probability that information will be discarded by reducing the amount of information of low frequency components processed by the interframe predictive coding circuit.

The low frequency component "L" separated by the coefficient separator 6 is encoded by a fully integral interframe predictive encoding circuit. This operation is as follows. Namely.

The low frequency components are quantized by a quantizer 7, further encoded by an encoder 17, and output to a multiplexer. The output signal of the quantizer 7 is simultaneously dequantized via the inverse quantizer 9 and then input to the adder 11, where it is added to the predicted value input from the coefficient frame memory 15 via the coefficient separator 16. A decoded signal is obtained.

The coefficient separator 16 has a boundary set at the same position as the coefficient separator 6, and sends only the low frequency components separated at the boundary out of the predicted values stored in the coefficient frame memory I5 to the adder 11. Send.

The output signal from the adder 1'l is stored in the coefficient frame memory 15 via the coefficient synthesizer 14, and is used as a predicted value for the next frame. Here, the coefficient synthesizer 14 is located at the same boundary line as the coefficient separator 6.

Coefficients are combined with a signal on the high frequency component element o side, which will be described later.

On the other hand, the high frequency component o separated by the coefficient separator 6 is
Basically, the same processing as the above-mentioned low frequency component reduction is performed in the leakage integral type interframe predictive coding circuit. However, in this case, since it is a leaky integral type, the decoded output signal from the adder 12 is multiplied by the prediction coefficient α in the prediction coefficient multiplier 13, and then combined with the lower frequency component in the coefficient synthesizer 14.
Thereafter, it is stored in the coefficient frame memory as a predicted value. By multiplying by this prediction coefficient α (0<α<l), even if an error or the like occurs, its influence will be gradually attenuated.

Therefore, even if this high frequency component u is temporarily discarded on the transmission path, it can be quickly reproduced on the receiving side.

In this way, the low frequency component element t is encoded by the fully integral interframe predictive encoding circuit, further passed through the encoder 17, and sent to the receiving side via the channel CH with a low discard rate of the multiplexer. On the other hand, the high frequency component H is encoded by a leakage integral type interframe predictive encoding circuit, further passed through the encoder 18, and sent to the receiving side via the channel CH2 of the multiplexer, which has a high discard rate.

As a result, when information is discarded due to congestion on the transmission path, the high frequency component "H" will be discarded first, and the low frequency component "L" will most likely be transmitted as is to the receiving side. Based on this low frequency component loss, the image can be reproduced even though it is unclear, and the situation where the 9-screen collapses can be avoided as much as possible.

On the other hand, when the congestion state of the transmission path is resolved, the high frequency component element O will reach the receiving side again, but this high frequency component element H will not be affected by the error (that is, the effect of the error will be Since the image is encoded using a leakage-integral interframe encoding method (which attenuates quickly), the screen related to the high frequency component u (that is, the portion related to the sharpness of the screen) can be quickly restored. Therefore, there is no need for a mode switching request from the receiving side to the transmitting side (there is no need for the sending side to take time to deal with discarding information, and it is possible to prevent the occurrence of an extreme freeze state of 9 screens).

Various modifications are possible in implementing the present invention. For example, in the above-described embodiment, discrete cosine transform was used as a method for orthogonally transforming the input image signal, but of course the present invention is not limited to this, and various orthogonal transform methods are applicable to the present invention. Also, low frequency component and high frequency component
The high-efficiency encoding method is not limited to the one in the embodiment (1) For example, the method for encoding the high frequency component "■" may be an intra-frame encoding method in which the influence of errors is quickly attenuated. However, it is also possible to use a leakage integral type interframe predictive coding method as a method for encoding low frequency components.

〔Effect of the invention〕

According to the present invention, even if information is discarded on the transmission path, the transmission of the low frequency component t of the two image signals to the receiving side is guaranteed with high probability, so blurring may occur. You can also avoid situations where the screen collapses. Also, the high frequency component I+ is encoded using a coding method that quickly attenuates the effects of errors, such as intraframe coding or leaky integral interframe predictive coding, so this high frequency component t+ is discarded. Even in such cases, the effects of errors caused by discarding information quickly attenuate, so there is no need to issue a request to the transmitter to switch to forced intra-frame encoding mode from the receiving example, and the occurrence of an extremely frozen state of three screens can be avoided. It can be prevented.

Moreover, in this case, depending on the properties of the two images, the low frequency component element
Since the separation ratio between l and low frequency components is controlled, the control is adaptively adapted to the characteristics of one image so that the encoding efficiency does not decrease and the distortion of the reproduced screen is prevented as much as possible. It is possible.

[Brief explanation of the drawing]

FIG. 1 is a principle block diagram according to the present invention. FIG. 2 is a block diagram showing an image encoding and transmitting apparatus as an embodiment of the present invention. FIG. 3 is a diagram for explaining blocks of transform coefficients, and FIG. 4 is a block diagram showing an image encoding and transmitting apparatus using a two-channel encoding and transmitting method. In fig. lj! Cosine transformer 2 Block line buffer 3 - Power calculation base 4 Controller 5 - Subtractor 6.16 - Coefficient numerator ILI base 7.8 Quantizer 9.10 - Inverse quantizer 11.12 - Addition unit 13 - pHjFl coefficient multiplier 14 - coefficient combiner 15 - coefficient frame memory 17, IEI - encoder

Claims (1)

[Claims] 1. Orthogonal transformation means (31) for orthogonally transforming an image signal into frequency components; power calculation means (32) for calculating the power distribution of the frequency components transformed by the orthogonal transformation means (31); separation means (33) for separating the frequency component transformed by the orthogonal transformation means (31) into a low frequency component and a high frequency component; ), a control means (34) for changing the separation ratio of low frequency components and high frequency components in the separation means (33), and a first encoding means for encoding the low frequency components separated by the separation means (33) with high encoding efficiency. (35), and a second encoding means (3) for encoding the high frequency components separated by the separating means (33) in a manner that allows fast attenuation of generated errors.
6) An image encoding and transmitting device comprising: and configured to transmit output signals of the first and second encoding means (35, 36) through two channels. 2. Orthogonal transformation means (31) for orthogonally transforming an image signal into frequency components; power calculation means (32) for calculating the power distribution of the frequency components transformed by the orthogonal transformation means (31); and orthogonal transformation means (31). ) separation means (33) that separates the frequency component converted by the converter into a low frequency component and a high frequency component; based on the power calculation result of the power calculation means (32), the separation means (33) separates the frequency component into a low frequency component and a high frequency component; and a control means (34) for changing the separation ratio of the high frequency components; a first encoding means (35) for encoding the low frequency components separated by the separation means (33) with high coding efficiency; A second encoding means (3
6), and having a channel with a low discard rate and a channel with a high discard rate, transmitting the low frequency signal encoded by the first encoding means (35) on the channel with a low discard rate, Means (36
2.) An image encoding and transmitting device comprising: multiplex transmission means (37) for transmitting a high frequency signal encoded in ) via a channel with a high discard rate.