JPH03263927A - Encoded output data quantity control system and decoder therefor - Google Patents

Abstract

PURPOSE:To surely control data quantity within a constant value with frame unit, etc., by coding control information representing by what method feedback control is performed with the same method as variable length encoding for an orthogonal transform coefficient, and transmitting it to a decoder side. CONSTITUTION:An accumulation value K outputted from a data quantity accumulator 31 is supplied to a quantization step shifter 32, and is converted to quantization step shift quantity DSq. The feedback control is applied to a predictive error for data quantity, and is controlled in detail in block unit. The control information required to be transmitted to the decoder side represents shift quantity in a quantization step that is the result of the feedback control. The data quantity can be remarkably reduced by performing runlength-coding on a part remaining unchanged. Since rough control can be performed in feedforward control, picture quality can be prevented from being remarkably fluctuated by the feedback control in block unit, and the capacity of a variable length buffer for data output can be also reduced.

Description

Detailed Description of the Invention (Industrial Application Field) This invention relates to a high-efficiency encoding method for efficiently encoding signals in fewer n months in recording, transmission, and display devices that process digital signals. This invention relates to a method for controlling the amount of encoded output data to make the amount of data uniform.

(Prior Art) Prediction residuals in predictive coding and orthogonal transform coefficients in transform coding have a considerably biased level distribution of generated signals.

Therefore, by assigning codes of different lengths depending on the occurrence of signal levels, the average value of the data amount can be made shorter than the fixed-length code.

The variable length code is determined by a Huffman code or the like, and the more biased the frequency of data occurrence, the lower the total amount of data.

By using variable length codes in this way, it is possible to reduce the average data size compared to fixed length codes while maintaining the same quality of the reproduced signal.

However, since the amount of data generated is not constant, in an actual device it is necessary to reduce the amount of data generated by 1IIJ'a to make it uniform.

The most common data amount IdJ m is feedback control using the data output buffer 7.

When the amount of encoded data increases and the amount of data i stored in the buffer increases, the quantization step is coarsened to reduce the amount of data.

Such a control method is used in systems such as video conferencing.

On the other hand, when considering the application of encoding to recording media, functions such as special playback are desired in addition to simple recording and playback functions.

Furthermore, in editing the encoded image data, the data length needs to be fixed in units of approximately one frame in order to replace the image data.

In such a case, with the feedback control, although the average value of the amount of data is kept constant, it is not possible to reliably keep the amount of data within a certain value in the desired unit.

Therefore, as a data amount control method that keeps the data amount within a certain fM, a previous application by the same applicant and the same inventor as the present application, Japanese Patent Application No. 1-14273 @r, "Control method for encoded output data amount" is proposed. be.

This method uses variable length coding and performs both feedforward and feedback control to ensure that the amount of data is kept within a constant mu without reducing coding efficiency or degrading image quality. .

(I!Ii problem that the invention seeks to solve) In conventional feedback control, high coding efficiency can be obtained by variable length coding, but it is not possible to reliably keep the amount of data within a certain value on a frame-by-frame basis. However, there was a problem in that it was difficult to apply to storage media.

On the other hand, the earlier patent application No. 1-14273 ``Control method for encoded output data amount'' controls high-efficiency encoding that performs normalization, and the data amount per block is controlled by a normalization coefficient. On the decoding side, using the transmitted normalization coefficients, the quantization is performed using the same processing as on the encoding side.
Since this method requires normalization processing, it cannot be applied to encoding without normalization, resulting in a lack of versatility.

The present invention has been made with attention to the above points, and the amount of data is controlled by both feedforward and feedback, and the control information on how the feedback i control is performed is based on the variable length of orthogonal transformation coefficients. Since it is encoded manually and transmitted to the decoding side in the same way as encoding, it is possible to reliably keep the amount of data within a certain value on a frame-by-frame basis, making it applicable to storage media, as well as special normalization processing, etc. The purpose of the present invention is to provide an unnecessary and versatile method for controlling the amount of encoded output data and a decoding device thereof.

(Means for Solving the Problems) In order to solve the above-mentioned problems, the present invention provides (1) data processing so that the amount of encoded output data for each predetermined period is within a certain value. .

In a method for controlling the amount of encoded output data that performs high-efficiency encoding, the amount of data is predicted in intervals smaller than the certain period, and the prediction in the certain period is performed based on the predicted data amount obtained. feedforward control means for performing encoding processing so that the total amount of data is constant; The present invention is characterized in that it is comprised of a feedback control means that performs one ITlIIll process, and a control information encoding means that encodes control information indicating what kind of control is performed by the feedforward and feedback control means and transmits it to the decoding side. 2. In the encoded output data amount control method according to claim (1), control information decoding means for decoding encoded and transmitted control information; The present invention provides a decoding device using an encoded output data amount control method, characterized in that it is comprised of encoded data decoding means that decodes encoded data under the control of ilJ m information.

(Operation) In order to solve the above problem, data is processed by both feedforward and feedback, and control information by feedback is variable length encoded and transmitted to the decoding side.

For data IIJIa, first, a predicted value of the data amount is obtained from the activity for each block, and a temporary quantization step is set by feedforward control.

Here, the activity uses the root mean square value of the AC component of the orthogonal transformation coefficient.

The difference between the predicted value of the data and the actual variable-length encoding is prevented from increasing by shifting the quantization step using feedback control on a block-by-block basis.

Control information indicating how feedback control is performed is encoded using a method similar to variable length encoding of orthogonal transform coefficients, and transmitted to the decoding side.

First, due to feedforward control, the amount of data becomes approximately the same as the target amount of data.

Furthermore, the error in the amount of data can be suppressed to about a few blocks by feedback processing on a block-by-block basis.

Although the fluctuation range of data i results in a loss with respect to the target data amount, this loss is extremely small.

Since feedback control is performed on data prediction errors, the image quality does not change significantly even with block-by-block control.

Since the feedback ilJ III information is encoded and transmitted, the amount of data increases accordingly, but it is small compared to the amount of data originally encoded.

As a result, the amount of data is fixed in desired units while performing variable length encoding.

(Embodiment) FIG. 1 is a block diagram showing an embodiment of an encoding apparatus using the encoded output data II/ILJ control method according to the present invention.

The image input signal is input from the image input terminal 1, supplied to the high-efficiency encoding processing section 2, subjected to high-efficiency encoding processing, and
An image output signal is taken out from the encoded data output terminal 3.

The high efficiency encoding processing section 2 further includes an orthogonal transformer 4, a coefficient memory 5, a quantizer 6, a variable length encoder 7, and a variable length buffer 8.

This high-efficiency encoding processing unit 2 is constructed by the International Telegraph and Telephone Consultative Committee CCITT, 5, except that a coefficient memory 5 is added.
Video conferencing method standardized in G15 (p x 6
4kbps codec).

While the processing unit in the earlier application, Japanese Patent Application No. 1-14273, requires coefficient normalization, this is common in that it does not require normalization, and can be said to be versatile.

The operation of the high-efficiency encoding processing section 2 will be explained below.

An image input signal is input from an image input terminal 1, supplied to an orthogonal transformer 4, and first converted into coefficients for each block of 8×8 pixels or the like.

As an orthogonal transform method, the discrete cosine transform (
DCT), etc., where one of the transform coefficients is a DC coefficient (average) and all other transform coefficients are AC coefficients (intersection).

Each coefficient, which is an output signal of the orthogonal transformer 4, is supplied to a coefficient memory 5.

The coefficient memory 5 is for compensating for a delay due to feedforward control, which will be described later, and is used to store a section 1ffi (here, for example, one frame) in which the amount of data is to be kept constant.
memory.

Since the number of coefficients is equal to the number of pixels before conversion, the coefficient memory 5
is similar to a frame memory, but the value after conversion is required to be 18 degrees higher than before conversion; for example, in the case of 8x8 DCT, about 12 pi] is required per coefficient.

The transform component, which is the output signal of the coefficient memory 5, is sent to the quantizer 6.
, and is uniformly quantized at an interval of 1 in a nesting step S (to be described later).

The output signal of the quantizer 6 is supplied to a variable length encoder 7.

Here, since most of the AC components are concentrated around O, they are converted into a preset Huffman code or the like. In this Huffman code, the word length of 0 is usually the shortest, and the word length increases as the absolute value of the level increases.

Furthermore, regarding O, instead of encoding each coefficient, the number of successive 0s is encoded.

The amount of output data obtained in this manner differs from block to block.

The output signal of the variable length encoder 7 is supplied to a variable length buffer 8.

The variable length buffer 8 is FIFO (First In
The input speed changes according to the amount of data output from the variable length encoder 7, but once it is stored and converted to a constant data rate, it is sent via the encoded data output terminal 3. High-efficiency encoded image encoded data is extracted.

Next, of the IIJm system of data amount, which is a characteristic part of the present invention, the feedforward IIIIII1 section 9 will be explained first.

The feedforward control section 9 further includes an activity detector 10, an activity memory 11, and a quantization step temporary setting device 12.

The feedforward ilJ m a [19 is for outputting a temporary staticization step PSq determined for each frame based on the activity of the orthogonal transform coefficient, and its operation will be explained below.

First, each coefficient, which is an output signal of the orthogonal transformer 4, is supplied to the activity detector 10.

The activity detector 10 obtains activity (activity degree) for each block from the conversion coefficients. This block does not necessarily have to be the same as the orthogonal transformation block, and several orthogonal transformation blocks may be grouped together.

Here, if this block is made larger, the amount of feedback control information data to be described later will be reduced, but fluctuations in the amount of data will also increase, resulting in a loss, so it is desirable that the block be large enough to balance the two.

Since the activity is used for predicting variable length encoded data, the higher the correlation with it, the better, and is given by the root mean square value of the AC coefficients. This value is not directly related to decoding, so if a better activity detection method is developed later, it can be replaced without causing compatibility issues.

The requested activity is in activity memory 1
1 and is supplied to the quantization step temporary setting device 12.

Activity method 11 is feedforward 1I
In order to compensate for the delay of ljIIl, the operation is delayed by one frame, and since the activity is for each block, the capacity is small compared to the coefficient memory 5 or the like.

The quantization step provisional setter 12 operates to obtain a provisional quantization step PSq such that the data reference for one frame is constant.

FIG. 2 is a block diagram showing an example of the configuration of the quantization step provisional setter in FIG. 1, and particularly shows an example of the configuration for parallel processing.

In FIG. 2, a plurality of data amount predictors, frame accumulators, and comparators are provided for each quantization step value, but the data amount predictor has a different number for each quantization step. Same process.

N types of quantization steps (for example, n is about 20) are set in advance by changing the value by about 10%.

The activity input when quantized at each quantization step is sent from the activity detector 10 via the activity input terminal 15 to each data amount predictor 16, 17 .
18, etc., and outputs prediction data 1PDtll, PDb2°, . . . pobn for each input activity.

This predicted value is an average value that is statistically determined in advance, and as the quantization step becomes finer, the amount of predicted data also increases. The value is written in the table ROM, and as soon as an activity is input, the predicted data amount is output.

The predicted data amount PD obtained in this way at the middle of the block
b1. PDb2. -PDbn is accumulated for one frame by the frame oneifter 19°20.21, and the predicted data amount for one frame at each quantization step is 01.02°.
...On is required.

DI, D2. ...On means that the data amount comparators 22 and 23 are turned on at the end of the frame for each quantization step.
．． At step 24, it is compared with the target data section, and it is determined whether it is smaller or larger than the target data section.

The result is input to the appropriate value determiner 25, which determines the finest quantization step among those smaller than the target data amount, and outputs the quantization step as a provisional quantization step PSa through the output terminal 26. are doing.

Since each input to the appropriate value determiner 250 is 1-bit information, the circuit can be configured with simple logic.

Next, the operation of the feedback control system 28 in FIG. 1 will be explained.

Based on the activity delayed during feedforward control in the activity memory 11 and the temporary quantization step PSa output from the quantization step temporary setting device 12, the data amount predictor 29 calculates predicted data IPD of each block.
get b.

The data predictor 29 is similar to the data 1 predictor 16.17.18 inside the ladder step temporary setting device shown in FIG.
．． The same thing as all 18 is combined into one.

The predicted data amount PDb is subtracted from the actual data portion Db selected from the variable length encoder 7 by subtraction 530 and is guided to the data amount accumulator 31.

In the data amount accumulator 31, (Db-PD
b) is added to the previous cumulative value 1fiK, and the new cumulative value K is held. The cumulative value is reset every frame.

The accumulated value output from the data amount accumulator 31 is supplied to the quantization step shifter 32 and converted into a quantization step shift amount 03 (l).

FIG. 3 is a diagram showing the quantization step shift i-characteristic.

The horizontal axis is the cumulative W4 value of the data amount error, and the vertical axis is the quantization step shift ID5a.

In FIG. 3, the maximum value of the cumulative value is set to WaX, and the minimum value is set to -laX, and this KlaX ultimately becomes the margin that must be provided for the feedback ilJ In.

The characteristic shown in FIG. 3 is such that the shift amount DSq increases as it approaches the edge so that K does not exceed this value under feedback control.

Temporary quantization step P temporarily set as shift ID 5Q
By adding S(l in an adder 33, the final quantization step SQ is obtained.

Since the predicted data IPDb is given a statistical average value, the integral value of the predicted data amount PDb is equal to the integral value of the actual data IDb in a boat fishing image, and K is close to O. Become.

However, if the actual data IDb is unevenly generated, K has a positive or negative value, and the quantization step SQ is shifted relative to the temporary quantization step PSa.

In this way, feedback IIJ II in the present invention
I is performed on the prediction error of the data part, and is finely controlled on a block-by-block basis.

If ilJ ill is performed correctly here, the data amount error for one frame at the end of one frame is K max
The predicted data amount in the provisional participation step PSq set in the feedforward system only needs to have a margin of K max with respect to the target data amount.

If K sax is the amount of data for 10 blocks, then the number of pixels in one frame is, for example, 720 x 480,
If the block size is 8×8, the total number is about 11,540, and the low efficiency of right-columnization can be almost ignored.

On the other hand, although such a control circuit is relatively complex, since the processing other than the activity detector 10 is performed on a block-by-block basis, the processing speed is only about 1/64 of the original signal.
Processing by a digital signal processor (digital signal processor) is also possible.

Next, the control information encoding means will be explained.

The control information that needs to be transmitted to the decoding side is the amount of shift of the quantization step that is the result of feedback IIJ m. Since the quantization step is set at approximately 10% intervals, the quantization step shift amount DSQ does not change for each block, but is considered to occur once every several blocks.

Therefore, by performing run-length encoding on portions that do not change, the amount of data can be significantly reduced.

FIG. 4 is a diagram showing an example of a change in the amount of quantization step shift. The horizontal axis is the interval of change (that is, the unit of time), and the vertical axis is the quantization step shift amount DS (l).

For example, if control as shown in FIG. 4 is performed, the information that needs to be transmitted is the interval of change and the degree of change.

Note that the change that occurs at one time is mostly ±1 step.

(In Figure 4, 1 step is 2%, so ±2
% change. ) For example, in FIG. 4, (6, +2) represents that the interval of change is 6 and the degree of change is +2%.

Therefore, the change interval is used as a run length, and the degree of change is encoded using an eight-man code or the like. This method is similar to orthogonal transform encoding. Processing is performed using a variable length code 1i134, and the encoded control information is sent to the control data output terminal 3.
5 and transmitted to the decoding side <ta device).

If the block size is 8 x 8 pixels and the quantization step changes once every 10 blocks on average, the data for one change will be about 6 bits, and the amount of data will be 0.02 bits per pixel.
It is about 01 bit.

Since the amount of normal data such as conversion coding is 1 to 2 bits per pixel, the data of this control information can be said to be sufficiently small.

Note that it is also necessary to send information on the temporary quantization step PSq, which is the result of the feedforward processing, but the type of PS(l) is about 20 and is several bits per frame, so the amount of data can be ignored.

FIG. 5 is a block diagram showing an embodiment of a decoding device using the encoded output data amount control system of the present invention.

In FIG. 5, the basic operation of the decoding means for the encoded data transmitted from the encoder tube is the same as the one described above.

First, since the image signal (encoded data input signal) inputted from the encoded data input terminal 37 has a different amount of data for each block, it enters the variable length buffer 38 and is taken out block by block.

Next, the data is converted from variable length to fixed length by a variable length decoder 39, and becomes data for each coefficient.

Each coefficient data is replaced by a representative value of each quantization in a quantization regenerator 40 and supplied to an inverse orthogonal transformer 41.

The inverse orthogonal transformer 41 performs the reverse processing of the orthogonal transformer 4 in FIG. 1 to obtain a reproduced image.

The obtained reproduced image is outputted via the reproduced image output terminal 42.

Next, the control information decoding means, which is a feature of the present invention, will be explained.

The present invention is characterized in that since quantization control information for each block is transmitted, quantization reproduction processing (inverse quantization) is controlled by this information.

Control information (iJ
J control data) is decoded by the variable length decoder 44, and the quantization step shift amount DSQ for each block is obtained.

DSq is added to the temporary quantization step PS(l, which is transmitted in advance at the beginning of the frame) by an adder 45, and becomes the actual quantization step Sq.

The quantization step Sq is supplied to the quantization regenerator 40, which performs a partial regeneration operation 111jIII to obtain a regeneration value.

With such an operation, correct reproduction values can be obtained even if quantization is controlled for each block.

Furthermore, in this method, since the control information for the quantization step is transmitted, a 1lIII information decoding means is required, but there is no need to provide a feedback control section in the decoding system (decoder W) as in the previous application.

As explained above, in the present invention, the amount of data is 1111 in both feedforward and feedback, and control information is variable-length encoded and transmitted, thereby improving image quality in general high-efficiency encoding using variable-length encoding. The amount of data can be kept constant with extremely high precision in the target section without deterioration or reduction in coding efficiency.

Rough control is performed using feedforward control, so there is no large fluctuation in image quality due to block-by-block feedback control, and there is no need to use a variable length buffer for data output. The capacity may also be small.

Since the amount of data can be predicted based on block activity, more accurate feedforward control is possible. There are no compatibility issues if you change the way activities are calculated later.

Since the main processing is performed in units of blocks, the processing speed may be quite slow, and it can be implemented even with a general-purpose DSP.

These characteristics make it possible to apply low-efficiency encoding processing using variable length encoding to storage media that requires special playback or data input in frame units.

(Effects of the Invention) The encoded output data amount control system and its decoding device of the present invention reduce the data amount by 1iIIIll by both feed forward and feedback, and by feedback III.
The ilJ If information about how @ is performed is encoded using the same method as variable-length encoding of orthogonal transform coefficients and transmitted to the decoding side, so the amount of data can be reliably kept within a certain value on a frame-by-frame basis. This method has extremely excellent practical effects, such as being able to be applied to storage media, and having versatility without the need for special normalization processing.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of an encoding apparatus of the present invention that controls the amount of encoded output data 1111, FIG. 2 is a block diagram showing an example of the failure of the quantization step temporary setting device in FIG. 1, Fig. 3 is a diagram showing the quantization step shift characteristics, Fig. 4 is a diagram showing an example of change in the quantization step shift amount, and Fig. 5
The figure is a block diagram showing an embodiment of a decoding device using the encoded output data amount control method according to the present invention. 1... Image input terminal, 2... High efficiency encoding processing section,
3... Encoded data output terminal, 4... Orthogonal transformer,
5...Coefficient memory, 6...Quantizer, 7.34...
・Variable length encoder, 8.38...Variable length buffer, 9.
... Feedforward control unit, 10... Activity detector, 11... Activity memory, 12.
...Quantization step temporary setting device, 15...Activity input terminal, 16.17.18.29...For predicting data amount, 19.20.21...Frame accumulator, 22,
23゜24...Comparator, 25...Appropriate value determiner, 2
6... Output terminal, 28... Feedback control section,
30...Reduction device, 31...Data-accumulator, 32.
...Quantization step shifter, 33...Adder, 35.
45...Control data output terminal, 37...Encoded data input terminal, 39°44...Variable length decoder, 40...
・Quantization regenerator, 41... Inverse orthogonal transformer, 42-...
Reproduction image output terminal, 43...l1IIll data input terminal, D1~On...predicted data amount for one frame,
Db... Actual data amount, DSQ... Quantization step shift amount, K... Cumulative value, IaX... Maximum value of cumulative value, PDb, PDbl ~ PDbn-.
- Predicted data amount, PSQ...temporary quantization step.

Claims (2)

[Claims] (1) In an encoded output data amount control method that performs highly efficient encoding by controlling the data amount so that the encoded output data amount for each predetermined period is within a certain value, Feedforward control means that predicts the data amount in units of intervals smaller than the fixed interval, and controls the encoding process based on the obtained predicted data amount so that the total predicted data amount in the fixed interval becomes constant; , a feedback control means for accumulating the difference between the predicted data amount and the actual encoded data amount and controlling the encoding process using this accumulated data amount; and what kind of control is performed by the feedforward and feedback control means. 1. A method for controlling the amount of encoded output data, characterized in that it is comprised of a control information encoding means that encodes control information about the operation performed and transmits it to the decoding side. (2) In the encoded output data amount control method according to claim (1), control information decoding means for decoding encoded and transmitted control information; 1. A decoding device using an encoded output data amount control system, characterized in that it is comprised of encoded data decoding means that performs a data decoding operation.