JPH04196790A - Quantizer - Google Patents

Abstract

PURPOSE:To suppress production of mosquito noise by using a small quantization width so as to attain detailed quantization when a DCT coefficient for an area sensitive to mosquito noise is quantized. CONSTITUTION:A value of eight or so is set to a switch 3 as a quantization width in which mosquito noise is not remarkable, and a selector control circuit 5 discriminates whether or not a DCT coefficient inputted from an input terminal 2 is resident in an area sensitive to mosquito noise and informs the result to a selector circuit 4. When the DCT coefficient inputted from the input terminal 2 is resident in an area sensitive to mosquito noise, the circuit 4 informs the setting value of the switch 3 to a quantization section 6, and when the DCT coefficient inputted from the input terminal 2 is not resident in an area sensitive to mosquito noise, the inputted quantization width inputted from the input terminal 1 is informed to the quantization section 6. Thus, the production of mosquito noise is suppressed.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The present invention relates to a quantization device used in a moving image encoding device such as a video telephone or a video conference system.

Conventional technology The development of video encoding technology has been remarkable, and together with the development of digital communication lines, international standardization activities are actively being carried out. 'J89 Image Coding Lecture p.
p, 43-48'', px64kbl)S
(p=1 to 30) A video encoding method is well known. Such a video encoding method performs discrete cosine transform (hereinafter, discrete cosine transform is abbreviated as DCT) on the interframe difference obtained by motion compensation prediction in block units of M pixels × N lines, and further The DCT coefficients are quantized by a quantization unit based on a quantization width g determined by the encoding bit rate, the amount of buffer remaining in the transmission buffer, and the like. Hereinafter, the value of the DCT coefficient after quantization will be referred to as a quantization representative value, and the value obtained by rounding coefficient i to an integer when the quantization representative value is decomposed into the product of quantization width and coefficient i will be referred to as quantization index. . Furthermore, in conventional video encoding systems, a quantization index, a quantization width g, block attribute information, etc. are variable-length encoded to generate a code string to be transmitted. At this time, the quantization index is encoded by scanning the quantization index in a zigzag scan order, and variable-length encoding the combination of a non-zero quantization index and a run length of zero preceding it.

The prior art of the above-mentioned quantization means will be explained below using FIG. 4 showing the quantization characteristics of the quantization section. In FIG. 4, the horizontal axis is the DCT coefficient before quantization, the vertical axis is the quantization representative value, and g is the quantization width. Overall, quantization errors are reduced by dividing the DCT coefficients into g units and using the central value as the quantization representative value. The region -gg is a region called a dead zone, where the quantization representative value for the DCT coefficient is set to zero. By quantizing a very small CT coefficient to a value O, the run length of 0 is increased and encoding efficiency is improved. As a result, the remaining buffer amount becomes a low level, so the value of the quantization width g for subsequent blocks becomes small, leading to an improvement in image quality.

Such a quantization unit takes the DCT coefficient and quantization width as address inputs, outputs the quantization representative value or quantization index, and has an internal ROM that stores the conversion table shown in Figure 4. , can be easily configured.

Problems to be Solved by the Invention In a video encoding system that combines motion compensation prediction and DCT, a noise called mosquito noise occurs in the boundary area or contour area between the moving area and the still area, which deteriorates the image quality of the reproduced image. It may cause deterioration.

This mosquito noise largely depends on the mid-range component of the DCT coefficient, and the mosquito noise becomes noticeable depending on the degree of noise superimposed on the mid-range component of the DCT coefficient. Furthermore, most of the mid-range components of the DCT coefficients have an amplitude of about 10, which is about the same as the average quantization width when encoding a moving image at a low bit rate such as 64 kbps.

In this way, the conventional quantization means described above quantizes the DCT coefficients within a block with a single quantization width.
In the middle range of the block, a large amount of noise is superimposed compared to the DCT coefficient value, and there is a problem in that mosquito noise appears conspicuously.

In view of the above problems, the present invention is sensitive to mosquito noise and attempts to suppress the mosquito noise when quantizing CT coefficients.

Means for Solving the Problems The present invention includes a selector circuit that selects an externally given first quantization width or a predetermined second quantization width depending on the position of a DCT coefficient; By having a quantization unit that quantizes the DCT coefficients using the quantization width,
This solves the problems of the prior art described above.

Operation The present invention attempts to suppress the occurrence of mosquito noise by using the above means to precisely quantize the DCT coefficients in a region sensitive to mosquito noise by using a small quantization width. It is something to do.

Embodiment FIG. 1 is a block diagram of a quantization device in an embodiment of the present invention. In FIG. 1, 1 is an input terminal for inputting the quantization width given from the outside, 2 is an input terminal for inputting DCT coefficients in accordance with the order of zigzag scan, and 3
4 is a selector circuit that receives the quantization width input from input terminal 1 and the set value of switch 3; 5 is a selector control that outputs a control signal to selector circuit 4; The circuit 6 is a quantizer which receives the output of the selector circuit 4 and the DCT coefficients input from the input terminal 2.

The operation of the above configuration will be explained below.

The switch 3 is set to a value of about 8 as a quantization width that makes mosquito noise less noticeable. The selector control circuit 5 determines whether the DCT coefficient input from the input terminal 2 is in a region sensitive to mosquito noise, and notifies the selector circuit 4 of the result. When the DCT coefficients input from the input terminal 2 are in the area sensitive to mosquito noise, the selector circuit 4 notifies the quantizer 6 of the setting value of the switch 3, so that the DCT coefficients input from the input terminal 2 If the area is not sensitive to mosquito noise, the quantization width input from the input terminal 1 is notified to the quantization unit 6. Similar to the prior art, the quantization unit 6 quantizes the DCT coefficients using a conversion table based on the characteristics shown in FIG.

The specific configuration of the selector control circuit 5 can be easily realized by a counter or the like that counts the clocks associated with the DCT coefficients if the area of the CT coefficients is determined to be sensitive to mosquito noise. Assuming that encoding is performed in blocks of 8 pixels x 8 lines, FIGS. 2 and 3 show examples of how to define regions of CT coefficients that are sensitive to mosquito noise.

In FIGS. 2 and 3, the shaded area is the area sensitive to mosquito noise. FIG. 2 is an example in which a region is defined in the middle range of two-dimensional spatial frequencies in accordance with the order of zigzag scanning. FIG. 3 is an example in which a region is defined by DCT coefficients having spatial frequencies only in the horizontal or vertical direction.

Note that in this example, D in the area sensitive to mosquito noise is
Although it has been described that the CT coefficients are quantized using a predetermined fixed value, it is also possible to quantize them using the smaller of an externally given quantization width and a predetermined quantization width.

Effects of the Invention As described above, according to the present invention, generation of mosquito noise can be suppressed by finely quantizing mid-range components of DCT coefficients, and the effect is significant.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a quantization device according to an embodiment of the present invention, and FIGS. 2 and 3 are diagrams showing how to determine areas of CT coefficients that are sensitive to mosquito noise in the same device.
FIG. 4 is a quantization characteristic diagram of a conventional quantizer. 4... Selector circuit, 5... Selector control circuit, 6
...Name of Quantization Department Agent Patent Attorney Haruaki Koebi and 2 others Figure 1 Figure 2 Figure 6 Figure 4

Claims (1)

[Claims] a switch for setting a predetermined second quantization width; a selector circuit for selecting an externally applied first quantization width or the second quantization width; and a selector control circuit for controlling the selector circuit. , and a quantization unit that quantizes information using a selected quantization width.