JPH0284884A - Picture encoding and transmitting system - Google Patents

Abstract

(57) [Abstract] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed Description of the Invention] [Summary] Regarding an image coding transmission method in which image signals are transmitted in bursts, an orthogonal transformation means for orthogonally transforming an image signal into frequency components and an orthogonal transformation means are used for the purpose of increasing encoding efficiency. ,
Separation means separates the frequency components transformed by the orthogonal transformation means into n groups according to the height of the frequency, and each frequency component separated by the separation means is weighted based on a predetermined prediction coefficient to calculate the difference. 1st to n
encoding means, and coefficient setting means for setting prediction coefficients according to the information discard state of the channel that transmits the output signal of the encoding means, the output signal of the first to nth encoding means is Configure to transmit using n channels.

[Industrial Field of Application] The present invention relates to an image coding transmission system that transmits image signals in bursts.

[Conventional technology]

In a high-efficiency image signal encoding device, the encoded image signal is transferred to an ATM (Asynchronous Tractor).
When transmitting using a burst multiplex transmission network such as nsfer mode) or packet switching, it is inevitable that transmission frames will be lost (cells and packets will be discarded) due to transmission errors, buffer overflows, and the like.

In a high-efficiency coding apparatus for image signals, interframe coding is performed on −C based on the difference between image frames, so if a transmission frame disappears, predictive coding becomes impossible. In this case, the receiving side generally issues a request to switch to forced intraframe encoding mode to the transmitting side, and the encoding method on the transmitting side is forcibly switched from interframe encoding to intraframe encoding, and the screen is The intraframe-encoded signal is sent to the receiving side, the screen is reproduced, and then the mode is switched to interframe-encoded mode again.

[Problem to be solved by the invention]

By the way, in the above-mentioned conventional method, since inter-frame predictive coding was performed, all signals after the transmission frame disappears become invalid, resulting in a problem that the screen becomes distorted. In addition, in order to return to the normal state, the receiving side needs to request the transmitting side to switch to forced intra-frame encoding mode, and in this case, forced intra-frame encoding has a lower encoding rate than inter-frame encoding. Due to the low conversion efficiency, the screen would temporarily freeze (fixation), resulting in a decrease in service quality.

As a method to deal with these drawbacks, a multi-channel coding transmission method (dated August 26, 1988 [Image The applicant has already proposed a coded transmission device.

FIG. 4 shows the configuration of an image encoding device using a two-channel encoding transmission method. In the figure, 4 ], 1 is an orthogonal transformer that orthogonally transforms the image signal into frequency components, 421 is a coefficient separator that separates the frequency components transformed by the orthogonal transformer 411 into low frequency components and high frequency components, 431 441 is a fully integral interframe encoder that encodes the low frequency components separated by the coefficient separator 421 with high encoding efficiency.
is a leaky integral type interframe encoder which encodes the high frequency components separated by the coefficient separator 421 in a manner in which the errors that occur are quickly attenuated. 451 has a channel CHI with a low discard rate and a channel CH2 with a high discard rate, and transmits the low frequency component encoded by the fully integral interframe encoder 431 through the channel CH1 with a low discard rate, and This is a multiplexing device that transmits the high frequency components encoded by the intercoder 441 through a channel CH2 with a high discard rate.

An input image signal is converted into frequency components by an orthogonal transformer 411, and further separated into a low frequency component t and a high frequency component by a coefficient separator 421, and each is encoded separately. In this case, the low frequency component [L contains information about the basic skeletal structure of the image, while the high frequency component o contains information about the fine parts of the image. Therefore, if low frequency components are lost due to information being discarded, etc., the reproduced image will be distorted, but even if only the high frequency components H are lost, the image will only become unclear; Not everything will collapse.

Therefore, the more important low frequency components are separately encoded by a fully integral type interframe encoder 431 with high encoding efficiency, and the high frequency components are encoded separately by a leaky integral type interframe encoder 441. The multiplexer 451 sends the signals to the receiving side via separate channels CHICH2.

The channel CHI is designed to have a low discard rate even when the transmission path is congested, and the probability that more important low frequency components are lost is low even when information is discarded on the transmission path.

Therefore, even if the high frequency component H via channel CH2, which has a higher discard rate, is lost due to information discard on the transmission path, the low frequency component H will be sent to the receiving side, so the image will be distorted on the receiving side. This can be prevented as much as possible.

In addition, since the high frequency component H is encoded using the leakage integral interframe coding method in which the errors that occur are quickly attenuated, even if the high frequency component H is lost in the transmission path, the normal high By sending the frequency component element, 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 transmission system, the prediction coefficients in the leaky integral interframe encoder 441 are determined with a predetermined drop rate in mind. The components are encoded inefficiently, resulting in poor encoding efficiency as a whole.

The present invention was created in view of these points, and an object of the present invention is to provide an image encoding and transmission method that improves encoding efficiency.

[Means to solve the problem]

FIG. 1 is a block diagram of the principle of the image coding transmission system of the present invention.

(i) In the invention diagram of claim 1, orthogonal transformation means 111 orthogonally transforms the image signal into frequency components.

Separation means 121 separates the frequency components transformed by orthogonal transformation means 111 into n frequency components from low frequency components to high frequency components.

The first encoding means 131 encodes the first low frequency component separated by the separating means 121.

The n-th encoding means 13n weights the n-th high frequency component separated by the separation means 121 based on a predetermined prediction coefficient, calculates a difference, and encodes it.

The coefficient setting means 151 sets a prediction coefficient according to the information discard state of the channel that transmits the output signals of the first encoding means 131 to the n-th encoding means 13n.

(ii) Invention of Claim 2 The image coding transmission system of Claim 2 has, in addition to the configuration of the image coding transmission system of Claim 1 described above, a channel with a low discard rate and a channel with a high discard rate. 1 encoding means 13
The first low frequency component signal encoded by 13n is transmitted through a channel with a low discard rate, and the nth high frequency component signal encoded by the nth encoding means 13n is transmitted through a channel with a high discard rate. The multiplex transmission means 161 is provided.

[For production]

(i) Invention of Claim 1 In the image coding transmission system of Claim 1, the orthogonal transformation means 111 converts the image signal into frequency components, and the separation means 12.1 converts the low frequency components into high frequency components. It is divided into n molecules 7 and encoded separately. in this case,
The low frequency component element I contains information about the basic skeletal structure of the image, while the high frequency component element 7 contains information about fine parts of the image. Therefore, the low frequency component +
etc. are lost, the playback screen will collapse, but
Even if the high frequency component element 7 is lost, it will only make the image unclear, and the entire image will not be destroyed.

Therefore, the first encoding means 131, such as a fully integral interframe encoder, encodes the low frequency component 1 with high coding efficiency, and also encodes the high frequency component 1 with a fast attenuation of generated errors. For example, the n-th encoding means 13n, such as a leakage-integral interframe encoder, encode the signals separately and send them to the receiving side via separate channels.

Further, in this n-th encoding means 13n, encoding is performed by weighting based on a predetermined prediction coefficient to obtain a difference, but this prediction coefficient is used depending on the information discarded state of the channel. The coefficient setting means 151 sets the coefficient accordingly.

In the invention of claim 1, the first to nth encoding means 1
By setting the prediction coefficients of 31 to 13n according to the information discard state of the channel, it is possible to perform optimal encoding according to the information discard state in each encoding means, and the overall encoding It can increase efficiency.

(ii) Invention of Claim 2 Further, in the image encoding transmission system of Claim 2, the n-th encoding means 13n converts the low frequency component signal encoded by the first encoder/stage 131 described above. The high frequency component signals encoded by the high frequency component signals are supplied to the multiplex transmission means 161, and the low frequency component signals are transmitted through channels with a low discard rate, and the high frequency component signals are transmitted through channels with a high discard rate. .

In the invention of claim 2, the high frequency component signal, which is less important than the low frequency component signal, is transmitted through a channel with a high discard rate, and each encoding means performs encoding according to the discard rate. Therefore, it is possible to eliminate wasteful coding and improve overall coding efficiency.

〔Example〕

Hereinafter, embodiments of the present invention will be described in detail based on the drawings.

FIG. 2 shows the configuration of an image encoding and transmitting apparatus in an embodiment in which the image encoding and transmitting method of the present invention is applied to a two-channel case.

In the figure, 211 is a cosine transformer that performs a two-dimensional discrete cosine transform (DCT) on the input signal in block units (for example, 8 x 8 pixels) to convert it into blocks of frequency component transform coefficients as shown in FIG. be. As shown in 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 coefficient separator 221 separates a block of transform coefficients transformed by the cosine transformer 211 into a low frequency component and a high frequency component. The separation method is, for example, the third
As shown in the figure, it is separated into low frequency components and high frequency components "□" at the boundary.

The separated low frequency component H is encoded by a complete integral type interframe predictive encoding circuit, while the separated high frequency component H is encoded by a leakage integral type interframe predictive encoding circuit.

This fully integral type interframe predictive coding circuit includes a quantizer 231. Inverse quantizer 232. Coefficient frame memory 233
．． Subtractor 234. It is composed of a closed circuit including an adder 235 and the like.

On the other hand, the leakage integral type interframe predictive coding circuit includes the quantizer 241 . Inverse quantizer 242. Coefficient frame memory 24
3. :$i calculator 244. Adder 245 Prediction coefficient multiplier 2
It consists of a closed circuit including 46 etc. The prediction coefficient multiplier 246 adds the prediction coefficient α(0<α
<1), and therefore the loop including this prediction coefficient multiplier 246 is a leaky integration type circuit.

The output signal (i.e., the output signal of the quantizer 231) from the fully integral interframe predictive coding circuit that encodes the low frequency component is inputted to the encoder 239 and converted into a code format according to the form of the transmission path. is encoded as .

In addition, the output signal of the leaky integral interframe predictive coding circuit that encodes the high frequency component □ (i.e., the quantizer 24
1 output signal) is input to the encoder 249 and encoded into a code format according to the transmission path configuration.

The output signals of these zero encoders 239 and 249 are sent to a multiplexer 261. The multiplexer 261 has two channels for signal transmission C)11. I have CH2. Channel CHI is a channel with a low discard rate of information discarded during information comparison on the transmission path, while channel C
H2 is a channel with a high drop rate. Here, the signal with the low frequency component is connected to the channel CHI with a low discard rate to prevent it from being discarded as much as possible, while the signal with the high frequency component x is connected to the channel CH2 with the high discard rate. When discarding information, it is discarded before low frequency components.

Further, 251 is a controller for setting the value of the prediction coefficient α of the prediction coefficient multiplier 246 according to the information discard state of the transmission path (channel 2). This controller 251 transmits information regarding the discard rate (the discard rate itself, a signal for specifying discard by the signal transmitting side, etc.) to the multiplexer 251.
61, and the prediction coefficient α is set according to this information.

When the discard rate is high, the value of the prediction coefficient α is reduced to control the influence of frequently discarded information to be quickly resolved. Furthermore, when the discard rate is low, the value of the prediction coefficient α is increased to perform control to perform more efficient encoding.

Next, the operation of the embodiment of the present invention will be explained.

The image signal is input to a cosine transformer 211, where it undergoes two-dimensional cosine transform on a block-by-block basis and is converted into frequency components. This result is supplied to the coefficient separator 221,
It is separated into a low frequency component and a high frequency component.

The low frequency components separated by the coefficient separator 221 are encoded by a complete component type interframe predictive encoding circuit. This operation is as follows. That is, the low frequency components are quantized by the quantizer 231 via the subtracter 234,
It is further encoded by an encoder 239 and output to a multiplexer 261.

The output signal of the quantizer 231 is simultaneously dequantized via the dequantizer 232 and then input to the adder 235, where it is added to the predicted value output from the coefficient frame memory 233 to obtain a decoded signal. This decoded signal is stored in the coefficient frame memory 233 as a predicted value of the next frame,
A subtracter 234 is provided. The subtracter 234 subtracts this predicted value from the low frequency component output from the coefficient separator 221 and supplies the subtraction result to the quantizer 231.

On the other hand, the high frequency component element separated by the coefficient separator 221,
Basically, the same processing as that for the low frequency component described above is performed on l by the leakage integral type interframe predictive coding circuit.

However, in this case, since it is a leakage integral type, the adder 24
The decoded output signal from No. 5 is multiplied by a prediction coefficient α by a prediction coefficient multiplier 246 and then stored in the coefficient frame memory 243.

By multiplying by this prediction coefficient α (0 × α × 1), the influence will gradually attenuate even when an error occurs or the transmission path is congested. Therefore, even if this high frequency component element H is temporarily discarded on the transmission path, it can be quickly reproduced on the receiving side.

Further, the value of this prediction coefficient α is adaptively set by the controller 251.
This prediction coefficient α is set according to the discard rate of channel CH2 sent from CH1, and when the discard rate is high (when the comparison state is frequent), the prediction coefficient α is set small; When the time is right, increase the prediction coefficient α.

In this way, the low frequency components are encoded by the fully integral interframe predictive encoding circuit, further passed through the encoder 239 and sent to the receiving side via the channel CHI of the multiplexer 261 with a low discard rate.

On the other hand, the high frequency components are encoded by a leaky integral type interframe predictive encoding circuit, further passed through the encoder 249, and sent to the receiving side via the channel CH2 of the multiplexer 261 with a high discard rate.

As a result, when information is discarded due to congestion on the transmission path, the high frequency component child o will be discarded first, and the probability that the low frequency component child will be transmitted as is to the receiving side is high. Therefore, on the receiving side, it is possible to reproduce an image, although it is unclear, based on this low frequency component, and it is possible to avoid the situation where the screen is distorted as much as possible.

On the other hand, when the congestion state of the transmission path is resolved, the high frequency component □ will reach the receiving side again, but this high frequency component ++ will not be affected by the error (in other words, the effect of the error will be Since the image is encoded using a leakage integral type interframe predictive coding method (which attenuates quickly), the screen related to the high frequency component o (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, and no time is required for the transmitting side to deal with information discard.

In this way, the low frequency component component separated by the coefficient separator 221 is encoded using the complete integral type interframe predictive coding method, and the high frequency component component n is encoded using the leaky integral type interframe predictive encoding method. By setting the prediction coefficient α in this leaky integral type encoding by the controller 251 according to the discard rate of the transmission path, it becomes possible to realize optimal encoding according to the discard rate, and the overall The encoding efficiency can be increased as follows.

Note that various modifications of the above-described present invention are possible. For example, in the above-described embodiment, discrete cosine transform was used as a method for orthogonally transforming the input image signal, but the present invention is not limited to this, and various orthogonal transforms are applicable to the present invention. In addition, the high-efficiency encoding method of low frequency component and high frequency component [8] is not limited to that of the embodiment. Alternatively, the high frequency component element may be encoded using another encoding method in which the influence of errors is attenuated according to a predetermined coefficient.

[Effects of the Invention] As described above, according to the invention of claim 1, predictive coding coefficients when each frequency component of an image signal is encoded by a plurality of encoding means are adjusted according to the information discard state of the channel. By setting these, it is possible to perform efficient encoding in each encoding means, and the overall encoding efficiency can be increased.

Further, according to the invention of claim 2, the higher the frequency component, the higher the discard rate, the higher the discard rate, the channel is transmitted, and the predictive coding coefficient is set according to the discard rate at this time. It becomes possible to perform encoding, and the overall encoding efficiency can be improved.

[Brief explanation of drawings]

Fig. 1 is a principle block diagram of the image coding transmission method of the present invention, Fig. 2 is a block diagram of an embodiment to which the image coding transmission method of the present invention is applied, and Fig. 3 is an explanatory diagram of blocks of transform coefficients. , FIG. 4 is an explanatory diagram of the two-channel coding transmission system. In the figure, 111 is an orthogonal transform means, 121 is a separation means, 131 is a first encoding means, 13n is an n-th encoding means, 151 is a coefficient setting means, 161 is a multiplex transmission means, 211 is a cosine transformer, 221 is a coefficient separator, 231.241 is a quantizer, 232.242 is an inverse quantizer, 233.243 is a coefficient frame memory, 234.244
is a subtracter, 235.245 is an adder, 246 is a prediction coefficient multiplier, 239.249 is an encoder, 251 is a controller, and 261 is a multiplexer. yang) 2 channels R of 1)

Claims (2)

[Claims] (1) Orthogonal transformation means (111) for orthogonally transforming an image signal into frequency components; and separation means (111) for separating the frequency components transformed by the orthogonal transformation means (111) into n groups according to the height of the frequency 121) and each frequency component separated by the separation means (121),
1st to nth encoding means (131 to 13
n); and coefficient setting means (151) for setting the prediction coefficients according to the information discard state of the channel that transmits the output signals of the first to nth encoding means (131 to 13n). , the first to n encoding means (131 to 13n)
An image coding transmission system characterized in that the output signal is transmitted through n channels. (2) orthogonal transformation means (111) for orthogonally transforming an image signal into frequency components; and separation means (111) for separating the frequency components transformed by the orthogonal transformation means (111) into n groups according to the height of the frequency. 121) and each frequency component separated by the separation means (121),
1st to nth encoding means (131 to 13
n); coefficient setting means (151) for setting the prediction coefficient according to the information discard state of the channel that transmits the output signals of the first to nth encoding means (131 to 13n); and a discard rate. The frequency component signals encoded by the first to nth encoding means (131 to 13n) are transmitted through a channel with a lower discard rate for lower frequency components, and are transmitted through channels with a lower discard rate for higher frequency components. An image coding transmission system characterized by comprising: multiplex transmission means (161) for transmitting on a high channel;