JPH04156794A - Encoding device - Google Patents

Abstract

PURPOSE:To perform accurate encoding with a small number of quantizers by performing the quantization while using the optimum quantizer, selecting a table combining the quantization means for each small block, and calculating the amount of data after encoding. CONSTITUTION:After the orthogonal transformation in a small block unit, the maximum value of the dynamic range of the orthogonal component other than the DC component after the orthogonal conversion for each small block or the absolute value is detected by a dynamic range detector 14, and the respective small blocks are classified according to the result of detection. The amount of data in a large block unit is calculated by a data amount calculator 25 from the obtained amount of data for each small block. In a quantization selector 26, the obtained amount of data in a large block unit and the preliminarily set amount of data are compared to decide the optimum width of the quantization. Further, the quantization of the orthogonal component stored in a buffer memory 27 is performed by a quantizer 28 based on the decided width of quantization, and variable encoding is performed by a variable encoder 29. Thus, accurate quantization can be performed with small number of quantizers 15 to 18.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of Industrial Application The present invention relates to an encoding device using orthogonal transformation.

2. Description of the Related Art Generally, the amount of information in an image signal is always large, so when recording on a digital recording/reproducing device such as a digital VTR, a method of reducing the amount of information by compression encoding is used as an effective means. Therefore, as an example of compression encoding technology, Hadamard transform and discrete cosine transform (DCT) are used.
There are some that use orthogonal transform encoding such as Orthogonal transform encoding encodes each frequency component obtained by frequency decomposing the input image signal into blocks, and encodes each frequency component obtained by dividing the input image signal into blocks. , the information amount of the original image signal can be reduced by techniques such as reducing the amount of data to be allocated.

Next, an encoding device using orthogonal transformation will be described.

FIG. 4 is a diagram showing a conventional encoding device, where 1 is an input terminal for sample values, 2 is a large block generator, 3 is a small block generator, 4 is an orthogonal transformer, and 5 is the amount of data. Calculator, 6 quantization selector, 7 buffer memory, 8 quantizer, 9
is a variable length encoder, and 10 is an output terminal.

The sample values of the image input from the input terminal l are divided into large blocks by the large block generator 2, further divided into rectangular blocks on the screen by the small block generator 3, and orthogonally transformed by the orthogonal transformer 4. be done.

Orthogonally transformed. The orthogonal components subjected to the orthogonal transformation are input to the data amount calculator 4 in units of large blocks, each consisting of a predetermined number of small blocks. The data amount calculator 5 calculates the amount of data after encoding in small block units for a plurality of quantizers prepared in advance, and based on the result, the quantization selector 6 calculates the amount of data after encoding for each small block. The converter is determined. At the same time, the orthogonal components input to the buffer memory 7 are quantized by the quantizer 8 using the quantizer selected by the quantization selector 6, variable-length encoded by the variable-length encoder 9, and output. It is output from terminal 10.

Problems to be Solved by the Invention However, the above-mentioned encoding device has the following problems. Encoding accuracy can be improved by having a large number of quantizers and finely switching the quantization width, but if there are m quantizers, the amount of data when quantization is performed in m ways for each block is needs to be calculated, and as the number of types of quantizers increases, the circuit size increases.

SUMMARY OF THE INVENTION In view of the problems of the prior art, it is an object of the present invention to provide an encoding device that can perform encoding with high accuracy using fewer types of quantizers.

Means for Solving the Problems The present invention provides a large block forming means for collecting sample values of an input signal and forming a large block, a small block forming means for dividing the large block into a plurality of small blocks, and a small block forming means for dividing the large block into a plurality of small blocks. orthogonal transformation means for performing orthogonal transformation for each block, m quantization means for quantizing orthogonal components obtained by the orthogonal transformation means with m types of quantization widths, and m quantization means for quantization means table selection means for selecting one quantization means table for each of the small blocks from among a plurality of types of quantization means tables consisting of n stages combining;
Data amount calculation means calculates the amount of data after encoding of each small block when quantized by each of the n-stage quantization means, and the data amount obtained by the data amount calculation means is used to calculate data. quantization means selection means for selecting an optimal quantization width for each of the small blocks, the amount of which is constant for each of the large blocks, and the orthogonal component using the quantization width selected by the quantization means selection means. This is an encoding device characterized by having a quantization means for quantizing, and an encoding means for encoding a quantized value obtained by the quantization means.

According to the above-described configuration, the present invention calculates the amount of encoded data in advance, quantizes it using an optimal quantizer, and creates a table consisting of n stages in which m quantization means are combined for each small block. , and calculate the amount of data after encoding.

Example Embodiments of the present invention will be described below with reference to the accompanying drawings.

FIG. 1 is a diagram showing an encoding device according to a first embodiment of the present invention, in which 10 is an input terminal for input sample values, 11 is a large block generator, 12 is a small block generator, and 13 is an orthogonal a converter; 14 is a dynamic range detector that detects the dynamic range of orthogonal components subjected to orthogonal transformation for each block;
15.16.17.18 are quantizers with different quantization widths,
19.20.21.22 is a quantizer 15 for each small block
．． 16, 17, and 18, a data amount calculator that calculates the amount of data when the quantized orthogonal components are encoded; and 23, which calculates the amount of data when the quantized orthogonal components are encoded; a quantizer table selector 24 that selects one table consisting of a combination of n quantizers from a table prepared in advance;
25 is a memory for storing the amount of data when the table selected by the table selector 23 is quantized and encoded by n quantizers; 25 is a data amount calculator that calculates the amount of data in large blocks; 26 is a quantizer selector that selects the optimal quantizer for each small block based on the code amount calculation result by the code amount calculator 25; 27 is a quantizer selector that delays orthogonal components until the quantizer is determined; 28 is a quantizer that quantizes orthogonal components to be actually transmitted, 29 is a variable length encoder, and 30 is an output terminal for variable length encoded data. The operation of this embodiment will be explained below.

After performing orthogonal transformation in units of small blocks, the dynamic range detector 14 detects the dynamic range or maximum absolute value of the orthogonal components excluding the DC component after orthogonal transformation for each small block, and calculates the maximum value according to the detection result. classifies each small block into classes. Hereinafter, the quantization method will be switched in small block units according to these classes, and the specific method will be described below.

First, the classes are classified into four types, and the classes are classified into four types in descending order of dynamic range: class a, class b, class C, and class d. Also, let us assume that the four quantizers 15°16 and 17.18 are Q1, Q2, Q3, and Q4, respectively, and Ql has the finest quantization width. do. Next, quantizers Q1, Q2, Q3, Q4
A basic table consisting of eight stages of quantization methods as shown in Table 1 is created by combining the following.

Table 1 In the basic table shown in Table 1, the smaller the number, the finer the quantizers are lined up. For actual quantization, Table 1
Based on the basic table of the dynamic range detector 14
Add the offset value according to the class found by
Four types of tables as shown in Table 2 are created, and each table is used to realize quantization.

Margin Table 2 Below In Table 2, the case of class d is the same as the basic table, and an offset value of 1 is added in the order of c, b, and a. In other words, small blocks with a small dynamic range are subjected to fine quantization, and small blocks with a large dynamic range are subjected to coarse quantization.

Next, the table selector 23 uses the data amount calculator 19 to determine the amount of encoded data when quantized based on Table 2, according to the classes already determined for all the small blocks in the large block. 20.21.22 and write the results to the memory 24. For example, if a large block consists of several small blocks, the memory 24 has 5ij in Table 3.
In Table 3, where the data amount of each small block is written as shown by (i: small block number, j: quantizer), the horizontal arrangement of small blocks and the class of each small block are as follows: The vertical direction represents the quantization stages shown in Table 2.

Table 3 The data amount calculator 25 calculates the data amount in large block units from the data amount in each small block obtained by the above method, and the quantization selector 26 calculates the data amount in large block units. The optimum quantization width is determined for each small block by comparing the amount with a preset data amount. Based on the determined quantization width, the quantizer 28 quantizes the orthogonal components stored in the buffer memory 27, and the variable length encoder 29 encodes the orthogonal components into variable length code.

As explained above, according to this embodiment, by selecting and encoding a table consisting of n stages combining m quantization means according to the dynamic range of a small block, fewer types of quantization means can be used. Good accuracy (quantization can be performed).

FIG. 2 is a diagram showing a second embodiment of the encoding device according to the present invention, and the basic components of FIG.
This embodiment is similar to the embodiment described above, but the configuration of the table selector 100 is different. That is, in this embodiment, when the input signal is a luminance signal and a chrominance signal, the offset is different when the class is classified according to the dynamic range.For example, when the input signal is a luminance signal, the table shown in Table 2 is used. , when the input signal is a color difference signal, the table shown in Table 4 is used.

In other words, the chrominance signal is quantized more coarsely than the luminance signal.

Table 4 As explained above, according to this embodiment, the deterioration of the color difference signal is less visually noticeable than that of the luminance signal, so the deterioration is less visually noticeable as a whole.

FIG. 3 is a diagram showing another method of dynamic range detection. In FIG. 3, the dynamic range detector 200 calculates the maximum dynamic range or absolute value of each small block before performing orthogonal transformation. Find the value.

Furthermore, in each embodiment, a table for each class was created by adding an offset to the basic table, but it is also possible to prepare several different tables in advance and select a table for each class from among them. .

Further, the number of classes, the number of quantizers, tables, etc. in each embodiment are merely examples, and can be set freely.

As described in detail, according to the present invention, it is possible to perform encoding with high precision using a small number of types of quantizers, and its practical effects are great.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of an encoding device according to a first embodiment of the present invention, and FIG. 2 is a block diagram of an encoding device according to a second embodiment of the present invention. 1
FIG. 3 is a block diagram showing another method of dynamic range detection, and FIG. 4 is a block diagram of a conventional encoding device. 13... Orthogonal transformer, 14,200...
・Dynamic range detector, 15.16.17.18
...Quantizer, 19°20, 21.22...
...Data amount calculator, 23.100...Table selector, 24...Memory, 25...
-Data amount calculator, 26...Quantization selector.

Claims (3)

[Claims] (1) A large block forming means that collects sample values of an input signal to form a large block, a small block forming means that divides the large block into a plurality of small blocks, and an orthogonal transform for each of the small blocks. orthogonal transformation means for performing the orthogonal transformation, m quantization means for quantizing the orthogonal components obtained by the orthogonal transformation means with m types of quantization widths, and a plurality of n stages in which the m quantization means are combined. quantization means table selection means for selecting one quantization means table from among the quantization means tables for each of the small blocks, and quantization using each quantization means of the selected quantization means tables; Optimal quantization in which the data amount is constant in units of the large block, using a data amount calculation means for calculating the data amount after encoding of each small block, and the data amount obtained by the data amount calculation means. quantization means selection means for selecting a width for each of the small blocks; quantization means for quantizing the orthogonal component using the quantization width selected by the quantization means selection means; 1. An encoding device comprising: encoding means for encoding a quantized value. (2) The quantization means table selection means selects the quantization means table according to the amplitude or dynamic range of the absolute value of each of the small blocks before or after orthogonal transformation. 1) The encoding device described. (3) The quantization means group selection means selects a quantization means table that is completely or partially different depending on whether the input signal is a luminance signal or a color difference signal. The encoding device described.