JPS63121321A - High efficiency encoding system - Google Patents

Abstract

PURPOSE:To adapt the distribution characteristic for variable-length encoding characteristic of a coefficient value and to prevent the decrease of compression effect of information quantity caused by unadaptation by switching plural variable-length code set and making a quantized coefficient variable-length encoded. CONSTITUTION:A quantization circuit 2 quantizes an orthogonal formation coefficient of a signal obtained by dividing an input picture into specified blocks. A sample reading circuit 4 sequentially outputs the synchronizing signals synchronizing signal generation circuit 3 for reading samples and the quantized conversion coefficient according to reading order. A variable-length encoding circuit 5 connected to the circuit 4 selects the variable length code corresponding to the quantized coefficient value from the specified variable-length code set and outputs it to a signal output terminal 8. The specification of the selection in terms of variable-length encoding is switched synchroneously with the switching of a variable-length code set storage circuit 7 by a variable-length code switching circuit 6, with a variable-length code set specification circuit 9 which actuates with the synchronizing signal of the circuit 3.

Description

DETAILED DESCRIPTION OF THE INVENTION (Technical field to which the invention pertains) The present invention relates to a high-efficiency encoding method for efficiently encoding audio and image signals.

(Conventional technology) In the following, encoding of an image signal will be explained as an example.

As a highly efficient encoding method for image signals, the image is divided into small blocks consisting of multiple samples, each of the small blocks is subjected to orthogonal transformation such as discrete cosine transformation, and the obtained orthogonal transformation coefficients are quantized. The quantized transform coefficients are 1
The one-dimensional signal is rearranged into dimensional signals, and the one-dimensional signal is sequentially variable-length coded. When all coefficients after a certain coefficient are zero, instead of giving variable-length codes to those coefficients, all coefficients after that coefficient are An orthogonal transform image coding method is known in which a code indicating that a coefficient is zero (block end code) is given, and the variable length code and block end code are multiplexed and transmitted. The orthogonal transform coefficients obtained by orthogonally transforming an image signal have different statistical properties between low-order coefficients and high-order coefficients, and the high-order components have a property that the signal power becomes smaller. For this reason,
For higher-order components, the probability that the quantized value becomes zero increases. Conventional methods utilize this property to collectively encode high-order component coefficients that are quantized to zero in the form of a block end code, thereby achieving an information compression effect.

(Problem to be Solved by the Invention) However, in the prior art described above, a single variable length code cell 1- is used for all coefficients before the program end code. Due to the difference in coefficient value distribution between low-frequency and high-frequency components, the characteristics of the variable-length code (Cera) do not match the quantized coefficient value distribution, making it impossible to obtain high coding efficiency. There was a drawback.

In addition, in the method using the block end code, the quantized value immediately before the block end code never becomes zero, so
The quantization value distribution of this coefficient is quite different from the quantization value distribution of coefficients of lower order than this coefficient. Therefore, there was a reduction in coding efficiency due to the use of a single variable length code set for coding coefficient quantization values with different distributions.

SUMMARY OF THE INVENTION An object of the present invention is to solve the problem of a decrease in the information compression effect caused by the mismatch between the distribution characteristics of coefficient values and the variable length encoding characteristics, and to provide an encoding method that achieves a high information compression effect.

(Structure of the Invention) The present invention switches a plurality of variable-length code sets in order to match coefficient value distribution characteristics and variable-length encoding characteristics.
The most important feature is variable length encoding of quantized coefficients. This method differs from conventional techniques in that a plurality of variable length code sets are used to match the distribution characteristics of coefficient values to be variable length coded.

(Implementation of the Invention I@) Example 1 FIG. 1 is a diagram showing an example in which the present invention is applied to a two-dimensional orthogonal transform encoding method for images, in which 1 is an input signal terminal;
2 is a quantization circuit, 3 is a synchronization signal generation circuit for sample readout, 4 is a sample readout circuit, 5 is a variable length encoding circuit, 6 is a variable length encoding set switching circuit, and 7 is a variable length code set storage circuit. , 8 is a signal output terminal, and 9 is a variable length code set designation circuit.

Orthogonal transform coefficients of a signal obtained by dividing an input image into blocks such as 4×4.8×8.16×16 are inputted from an input terminal 1 and quantized by an ff1r conversion circuit 2. The sample reading circuit 4 sequentially outputs quantized orthogonal transform coefficient values in accordance with the synchronizing signal from the sample reading synchronizing signal generating circuit 3 and the reading order shown in FIG. FIG. 3 is an example of a table for specifying a variable length code cell I~ for each transform coefficient, and shows the number of the variable length code set corresponding to the transform coefficient. The variable length code set designation circuit assigns a variable length code set number for each coefficient according to the table shown in FIG.
Output to. The variable length encoding circuit 5 selects a variable length code corresponding to the quantized coefficient value from the specified variable length code set and outputs it to the signal output terminal 8.

This embodiment focuses on the fact that the quantization value distributions of low-order components and high-order components of orthogonal transform coefficients of an image are different, and is an example in which encoding efficiency is improved by the present invention. In the structure of this embodiment, a variable length code set suitable for the quantization value distribution of each orthogonal transform coefficient can be used, so efficient encoding is possible. As a result, even when obtaining a reproduced image of the same quality, the method requires less information than the conventional method.

Embodiment 2 FIG. 4 is a diagram showing an embodiment in which the method described in claim (3) of the present application is applied to a two-dimensional orthogonal transform encoding method for images, and 101 is a signal input terminal. , 102
104 is a quantization circuit, 104 is a sample reading circuit, 105 is a variable length encoding circuit, 109 is a final non-zero quantization coefficient detection circuit, 107 is a variable length code cell 1 to storage circuit, 106 is a variable length code set switching circuit , 105 is a variable length encoding circuit, 11O is a block end code generation circuit, 111 is a signal multiplexing circuit, and ID8 is a signal output terminal.

The quantization circuit 102 and the sample reading circuit 104 are of the first embodiment.
It is similar to Final non-zero quantization coefficient detection circuit! 09 is
By monitoring the quantized values of the one-dimensional orthogonal transform coefficients by the sample reading circuit 104, it is possible to detect which coefficient the quantized values of all coefficients after which are zero, and to detect the coefficients immediately before that coefficient. The position of is output as the final non-zero quantized coefficient position. The variable length code set switching circuit 106 normally selects the variable length code set 1, but when variable length coding the final non-zero quantized coefficient, the variable length code set switching circuit 106 selects the variable length code set 1-1.
Select 2. The variable length encoding circuit 105 performs variable length encoding on the quantized coefficients using the variable length code set selected by the variable length code set switching circuit 106. The block end code generation circuit 110 encodes the final Immediately after the non-zero quantized coefficient, a block final code indicating that all subsequent quantized values are zero is generated. Multiplexing circuit 11
1 multiplexes the variable length code and the block end code and outputs the multiplexed code to the signal output terminal 108. Here, the sets of binary patterns of variable length codes included in variable length code set 1 and variable length code set 2 are the same, and only the correspondence between quantized values and binary patterns differs. When decoding this code, normally the quantized value is reproduced from the variable length code based on variable length code set 1, but when a block end code is detected, the quantized value reproduced immediately before is canceled. , regenerate the quantized value immediately before the block end code based on variable length code set 2.

This embodiment focuses on the fact that the quantization value distribution of orthogonal transform coefficients of an image signal block is different between the final non-zero quantization coefficient and the previous quantization coefficient values, and uses the present invention to improve coding efficiency. This is an example.

In the structure of this embodiment, it is possible to use a variable length code set for the final non-zero quantized coefficient that has different characteristics from the variable length code set for the quantized coefficients before the final non-zero coefficient. Therefore, by switching between two variable-length code sets that match the respective quantization value distributions, coding efficiency is improved. The effect is that even when obtaining a reproduced image of the same quality, there is less information compared to the conventional method. There was an improvement in that only 4affl was required.

(Effects of the Invention) As described above, there is an advantage that encoding efficiency is improved by switching a plurality of variable-length code sets to match the distribution of quantized values to be variable-length encoded. As a result, there is an improvement in that even if a reproduced image of the same quality is obtained, a smaller amount of data is required compared to the conventional method.

[Brief explanation of the drawing]

FIG. 1 is a diagram illustrating an image orthogonal transform encoding system having a function of adaptively switching between multiple variable-length codes according to the present invention, and FIG. 2 is a sample reading circuit of the first embodiment. Figure 3 shows the orthogonal transform coefficient reading order of the first embodiment. FIG. 4 is a diagram showing another embodiment of the present invention. In the figure, 2.102...Quantization circuit, 3...Synchronization signal generation circuit for sample reading, 4.104
. . . sample reading circuit, 5, 105 . . . variable length encoding circuit, 6.106 . . . variable length code set switching circuit, 7.
107...Variable length code set storage circuit, 9...Variable length code set designation circuit, 109...Final non-zero quantization coefficient detection circuit, +10...
・The block end code generation circuit is shown. Conversion coefficient order (vertical)

Claims (4)

[Claims] (1) Means for quantizing all or some samples in a block composed of a plurality of samples of various types of discretized signals, and means for variable length encoding the quantized sample values. , means for switching a variable length code set indicating a correspondence relationship between a quantized value and a variable length code for each sample value or for each plurality of sample values; 1. A high-efficiency encoding method, comprising: means for variable-length encoding the sampled values. (2) In the high-efficiency encoding method, the quantized sample values are sequentially scanned, and the quantized sample values read out by the scanning are all performed in a certain specific manner from a certain sample to the final sample of the block. means for detecting which sample after which all the samples are the specific value, and a means for detecting which sample after the detected sample is quantized to the specific value until the final quantized sample of the block. A high-efficiency encoding system according to claim 1, further comprising means for generating a block end code indicating that the block has ended. (3) Variable-length coding set that variable-length encodes the quantized value of the final non-zero sample, which is the sample immediately before the block end code, and variable-length encoded the quantized value of the sample other than the final non-zero sample. The high-efficiency encoding system according to claim (2), further comprising means for switching to a variable-length code set different from the variable-length code set used in the present invention. (4) without changing the binary pattern of the variable length code of the variable length code set for samples other than the final non-zero sample;
Claim 1, characterized in that a variable length code set configured by changing only the correspondence between the value pattern and the quantized value is used for variable length encoding of the quantized value of the final non-zero sample. High-efficiency encoding method described in section 3).