JPH05160745A - Quantizer - Google Patents

Abstract

PURPOSE:To provide a quantizer which hardly causes a non-coding block for a block possibly causing picture quality deterioration if it becomes a non-coding block. CONSTITUTION:A constant (a) is set to a constant number setting device 3, the constant (a) are quantized by a quantizer 5 at the same quantizing width as that of a quantizer 4. On the other hand, a quantizer discrimination device 8 discriminates whether a processed block is a non-coding block or a coding block by monitoring an output of the quantizer 4 to control a selector 9. Then when the processing block is the coding block, the selector 9 selectively outputs an output of the quantizer 4 and when the processing block is the non-coding block, the selector 9 selectively outputs an output of the quantizer 5 to an output terminal 10.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a moving picture coding apparatus such as a videophone, a video conference system, and a multimedia system, and more particularly to a quantizing apparatus for the moving picture coding apparatus using orthogonal transformation.

[0002]

2. Description of the Related Art In recent years, the development of moving picture coding technology has been remarkable, and international standardization activities have been actively carried out. Among them, "CCIT Rec (CCITT Rec
orientation H.M. 261-Video co
dec for audiovisual service
ces at p × 64 kbits / s) ”is well known. Such a moving image coding method is based on aM pixels × bN lines (a, b,
(M and N are natural numbers) The inter-frame difference obtained by motion compensation prediction in units of M pixels × N lines is subjected to discrete cosine transform (hereinafter, discrete cosine transform is abbreviated as DCT), and the coding bit rate and The quantization width is determined from the buffer remaining amount in the transmission buffer, and the DCT coefficient is quantized by the linear quantizer. Below, DC after quantization
The coefficient when the value of the T coefficient is decomposed into the product of the quantization width and the coefficient is rounded to an integer and is referred to as a quantization index.

Further, in the conventional moving image coding system, the quantization index is scanned in a zigzag scan order,
A combination of a non-zero quantization index and a preceding run length of 0 is variable-length coded, and attribute information is further added to generate a code string to be transmitted. The attribute information includes discrimination between intra-frame / inter-frame coding, presence / absence of motion compensation prediction, discrimination between coded / non-coded blocks, motion vector, and quantization width. All quantization indexes are 0
Blocks are called non-coded blocks, and blocks that are not non-coded blocks are called coded blocks. The distinction between coded / non-coded blocks is that motion compensation is performed when the non-coded blocks are decoded. The block image at the corresponding position of the frame used for prediction is transmitted so that it can be used as it is as the image of the block.

[0004]

However, the difference between frames is small in a region in which an object having a small change in luminance from the background is moving, or a region in which an object having a fine pattern is moving, and therefore all of the blocks in the block are DCT
The coefficient may not be larger than the value of the quantization width. In such a case, in the above-mentioned conventional technique, since the block becomes an uncoded block and the motion of the image is not preserved, the image quality is deteriorated due to missing objects or appearance of non-existing objects. Had challenges.

The present invention solves the above-mentioned conventional problems, and an object of the present invention is to provide a quantizer which is less likely to become a non-encoded block in which a non-encoded block leads to image quality deterioration.

[0006]

In order to achieve this object, the quantizing means of the present invention comprises first and second quantizers having the same quantizing characteristics as those of the prior art, and a second quantizing means. It has a constant setter for setting the value of the quantization width input to the input unit and a quantizer decision unit for monitoring the output of the first quantizer.

[0007]

With this configuration, the non-coding block is detected by the quantization decision unit, and in the case of the non-coding block, fine quantization is performed by the second quantizer, so that the non-coding block leading to image quality deterioration You can reduce the block.

[0008]

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

FIG. 1 is a block diagram of a quantizer according to an embodiment of the present invention. In FIG. 1, 1 is DCT
An input terminal for inputting a coefficient, 2 an input terminal for inputting a quantization width, 3 a constant setter for setting a value of a predetermined quantization width, 4 a first quantizer, 5 a second quantum The first quantizer 4, the first buffer 7 having the output of the first quantizer 4 as an input,
Is a second buffer having the output of the second quantizer 5 as an input, 8 is a quantization decision unit having the output of the first quantizer 4 as an input, and 9 is the first and second quantizers. Of the quantization decision device 8 as an input to the control signal input, and
0 is an output terminal for outputting the output of the selector 9 to the outside, 11
Is an output terminal for outputting the determination result of the quantization determiner 8 to the outside, and 12 is an output terminal for outputting the output of the constant setting device 3 to the outside.

The operation of the above arrangement will be described below. The first quantizer 4 equivalent to the conventional technique quantizes the DCT coefficient given from the input terminal 1 with the quantization width given from the input terminal 2. First buffer 6
Temporarily stores the output of the first quantizer 4.

In the constant setter 3, a predetermined constant a is set as the quantization width. The constant a is a constant determined in consideration of the coding bit rate and the like in order to maintain a certain limit of image quality, and is set to a smaller value as the coding bit rate is higher. The second quantizer 5 is a quantizer having the same quantizing characteristics as the first quantizer 4, and the D quantizer inputted from the input terminal 1
The CT coefficient is quantized by the output of the constant setter 3. The second buffer 7 temporarily stores the output of the second quantizer 5.
Further, since the value of the constant a is also necessary for accurate decoding, the value of the constant a is output to the outside from the output terminal 12 so that it can be used as attribute information.

The quantization decision unit 8 decides whether the block is an uncoded block or a coded block by monitoring the output of the first quantizer 4, and notifies the selector 9 of the decision result. At the same time, the output terminal 11 notifies the outside so that it can be used as attribute information. Selector 9
Switches the output of the quantizer determiner 8 as a control signal. That is, when the block is a coded block, the information output from the first buffer 6 is output, and when the block is a non-coded block, the second block is output.
The information output by the buffer 7 of is output. The information temporarily stored in the first buffer 6 and the second buffer 7 is
It is read in synchronization with the clock supplied after the determination by the quantization determiner 8 and is output to the outside from the output terminal 10 via the selector 9.

Quantization using the second quantizer 5 may result in non-coded blocks, but in that case, a significant quantization index to be transmitted is almost always present, such as a stationary background. This is a case where there is no such case, and even if the block image at the corresponding position of the frame used for the motion compensation prediction is used as it is as the image of the block, the image quality is not deteriorated.

First quantizer 4 and second quantizer 5
Can be easily configured by having a DCT coefficient and a quantization width as an address input, a quantization index as an output, and a ROM having a predetermined conversion table stored therein. The quantization decision unit 8 can be easily configured by a comparator that compares the output of the first quantizer 4 with the value 0, a counter that counts the decision result of the comparator, and the like.

As described above, according to the present embodiment, since the DCT coefficient having the same value as the input of the first quantizer 4 is quantized with the predetermined quantization width, the first quantizer 4 Blocks that become uncoded blocks when quantized
It can be changed to a coded block.

In this embodiment, two quantizers having the same quantization characteristic are used. However, when it is determined that the block is an uncoded block, the DCT coefficient is multiplied by a and coded again. In this case, the first quantizer and the second quantizer can be realized by one quantizer.

[0017]

As described above, according to the present invention, the first and second quantizers having equivalent quantization characteristics, the constant setter for setting the quantization width given to the second quantizer, By providing a quantizer determiner that monitors the output of the first quantizer, it is possible to change a non-coded block from a non-coded block to a coded block for coding, so that a moving image with a small luminance change can be coded. A good image with no missing or overlapping is obtained, and the effect is great.

[Brief description of drawings]

FIG. 1 is a block diagram of a quantizer according to an embodiment of the present invention.

[Explanation of symbols]

 3 constant setter 4 first quantizer 5 second quantizer 8 quantization determiner 9 selector

Claims (1)

[Claims] 1. A first quantizer for quantizing an input signal with an externally given quantization width, a constant setter for setting a constant, and an input signal for quantizing with an output of the constant setter. The outputs of the second quantizer and the first quantizer are all 0
A quantizer including a quantizer determiner for determining whether or not the selector and a selector for switching the outputs of the first and second quantizers to one according to a determination result of the quantizer determiner.