JPH05219495A - Quantization characteristic switching processing method - Google Patents

Abstract

PURPOSE:To prevent a transmission buffer memory from being turned to an under flow or an over flow by changing a quantization characteristic based on the inclination of the increase and decrease of the transmission buffer memory. CONSTITUTION:In an encoding processing part 100, a difference between an inputted video and a predicted value is searched, a discrete cosine conversion is operated, and the data are quantized, and supplied to a transmission buffer memory 7. Then, the content of the transmission buffer memory 7 is outputted to a transmission path. In a quantization characteristic control part 8, the inclination of the change of the using amounts of the transmission buffer memory 7 is inspected, and the quantization characteristic is controlled. That is, at the time of operating the control of the quantization characteristic based on the using amounts of the transmission buffer memory 7, the control is implemented based on the inclination of the change of the using amounts of the transmission buffer memory 7. Thus, the rapid change of the using amounts of the transmission buffer memory 7 can be hardly generated, so that the rapid change of a picture quality can be prevented.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a quantization characteristic switching processing method in video coding processing in which a communication information amount is reduced and transmitted during video communication.

[0002]

2. Description of the Related Art Not only in transform coding, but in a coding device in which information to be transmitted is not constantly generated, a transmission buffer memory is provided in order to match with a transmission line for transmitting information at a constant speed. However, information generation is controlled so that the transmission buffer memory does not underflow or overflow. To avoid underflow, make the quantization step size fine so that more information will be generated when the amount of transmission buffer memory usage falls below a certain value, while coarsen the quantization step size to avoid overflow. ..

[0003]

Although the transmitted image is transmitted and received between the transmitting and receiving devices by such a device, the characteristic of the quantizing unit is determined according to the amount of use of the buffer memory at every moment, so that the characteristic of the image is changed. Image quality is often degraded without consideration.

An object of the present invention is to control the quantization characteristic so that the transmission buffer memory will not underflow or overflow.

[0005]

FIG. 1 is a block diagram showing the principle of the present invention. In the figure, reference numeral 2 is an A / D conversion unit, 7 is a transmission buffer memory, 8 is a quantization characteristic control unit, and 100 is an encoding processing unit.

In the encoding processing unit 100, the difference between the input image and the predicted value is obtained, and then the discrete cosine transform is performed, and the image is quantized and supplied to the transmission buffer memory 7. Then, the contents of the transmission buffer memory 7 are output to the transmission path side.

In the quantization characteristic control unit 8, the change tendency of the amount used in the transmission buffer memory 7 is checked to control the quantization characteristic.

[0008]

In the case of the present invention, in the conventional control of the quantization characteristic based on the usage amount in the transmission buffer memory 7, the control is performed by observing the change tendency of the usage amount in the transmission buffer memory 7. To be done.

As a result, the amount of use of the transmission buffer memory 7 is unlikely to change abruptly, and the image quality can be prevented from changing abruptly.

[0010]

FIG. 2 shows an embodiment of the present invention for processing an image block of 8 × 8 pixels. In the figure, 1 is a video input terminal, 2 is A
/ D conversion unit, 3 subtraction unit, 4 prediction value generation unit, 5 image block conversion processing unit, 6 quantization unit, 7 transmission buffer memory, 8 quantization characteristic control unit, 9 transmission line, Reference numeral 10 is an inverse quantization unit, 11 is an inverse transformation processing unit, and 12 is an addition unit.

The image added to the input terminal 1 is digitized by the A / D converter 2, and the subtracter 3 obtains the difference from the output of the predicted value generator 4. Conversion processing unit 5
In order to execute the discrete cosine transform (DCT), the input digital signal is converted into a block of 8 lines × 8 pixels, and then the discrete cosine transform is performed.

The cosine transformed signal (hereinafter referred to as transform coefficient) is divided by a quantizing coefficient suitable for each transform coefficient in the quantizing unit 6. Here, the quantization coefficient used in the quantization unit 6 uses the quantization characteristic determined by the quantization characteristic control unit 8 based on the usage amount of the transmission buffer memory 7.

The quantized transform coefficient is sent to the transmission line 9 via the transmission buffer memory 7, and the prediction value is generated in the addition unit 12 via the inverse quantization unit 10 and the inverse transformation processing unit 11. After being added to the output of the unit 4, it is input to the prediction value generation unit 4 and used for generation of prediction values for subsequent signals.

FIG. 3 shows an example of a circuit of the quantization characteristic control section of FIG. Reference numeral 13 is a transmission buffer memory amount display signal which is supplied from the transmission buffer memory 7, 14 is a shift register, 15 is a clock signal, 16 is a subtracting section, 17
Is a double multiplication unit, 18 is a 3-input addition / subtraction unit, 19 is a characteristic determination circuit, and 20 is a quantization characteristic output.

The transmission buffer memory amount display signal 13 is input to the shift register 14. The contents of the shift register are shifted by the clock signal 15 having a pulse for each processing of the block, the first-order differential value of the used amount (that is, the difference between adjacent signals) is output from the subtraction unit 16, and the second-order differential value ( That is, the addition / subtraction value obtained by weighting “1” vs. “−2” vs. “1” between adjacent three signal values is 2 times the multiplication units 17 and 3.
It is output via the input addition / subtraction unit 18.

These outputs and the transmission buffer memory amount display signal are input to the characteristic determining circuit 19. In the characteristic determination circuit 19, one is selected from n quantization characteristics (n is determined by the used transmission buffer memory amount) depending on the transmission buffer memory amount (this is called n _i ). This value is obtained by dividing the maximum quantized coefficient and the minimum quantized coefficient into n equal parts. Next, n _i is calibrated by the positive, 0, and negative of the first-order differential value and the second-order differential value.

For example, if both are positive, 2/3 of the difference between levels is added to n _i . Both in the case of negative reduces the 2/3 interlevel difference from n _i. First derivative is positive, if second derivative is negative add 1/3 interlevel difference n _i. Conversely, 1
Negative floors derivative, second derivative is reduced to 1/3 of the inter-level difference from n _i in the case of a positive. When the first derivative is 0, n
_{When i} is used as it is and the second derivative is 0, the control is based on the first derivative. This control result is sent to the quantizer as the quantization characteristic output 20.

[0018]

According to the present invention, since the quantization characteristic is changed based on the increasing tendency and the decreasing tendency of the transmission buffer memory, it is difficult for the buffer memory usage amount to change abruptly. Is smoothed, and rapid changes in image quality can be prevented.

The output of the clock pulse is not limited to the block unit, and the degree of smoothing of the change in the quantization characteristic can be controlled by generating it in accordance with a plurality of block processing timings. It is possible to easily meet the conditions of.

[Brief description of drawings]

FIG. 1 shows a principle configuration diagram of the present invention.

FIG. 2 shows an embodiment of the present invention.

FIG. 3 shows a circuit example of a quantization characteristic control unit shown in FIG.

[Explanation of symbols]

 1 Video Input Terminal 2 A / D Conversion Section 3 Subtraction Section 4 Prediction Value Generation Section 5 Image Block Conversion Processing Section 6 Quantization Section 7 Transmission Buffer Memory 8 Quantization Characteristic Control Section 9 Transmission Line 10 Inverse Quantization Section 11 Inverse Transformation Processing Section 12 adder section 13 transmission buffer memory amount display signal 14 shift register 15 clock signal 16 subtraction section 17 double multiplication section 18 3 input addition / subtraction section 19 characteristic determination circuit 20 quantization characteristic output 100 encoding processing section

Claims (1)

[Claims] 1. A transform coding method in which an input image signal is divided into blocks of a fixed number of pixels, the block is transformed using an orthogonal transformation matrix, and each signal in the transformed block is quantized and encoded. The quantization characteristic used for quantization is selected and used from a preset quantization characteristic group according to the amount of change in the amount of transmission buffer memory used for smoothing information transmission within a certain period of time. Quantization characteristic switching processing method.