JPH06284407A - Encoder - Google Patents

Abstract

PURPOSE:To obtain the encoder whose information quantity is allocated depending on the movement of a picture. CONSTITUTION:Picture data subjected to block processing and orthogonal transformation are classified by a classifying device 12 and one of quantizers 5a-5d whose quantization step differs is selected depending on the classification result and the data are fed to a variable length encoder 9. Moreover, a motion detector 11 detects a motion of the picture data subjected to block processing and revises the classification result by the device 12 depending on the detected motion.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a coding device for coding an image signal using an orthogonal transform coding method.

[0002]

2. Description of the Related Art Conventionally, various encoding methods have been proposed in order to reduce the amount of transmission data when digitally transmitting or digitally recording information such as images. A transform coding method is known in which an image to be transmitted to one of them is orthogonally transformed, and the transformed data is quantized. A method is also known in which the quantization table is switched according to the amount of information in the image. On the other hand, after the above quantization,
There is known a variable length coding method in which a short codeword is assigned to a code having a high occurrence frequency and a long codeword is assigned to a code having a low occurrence frequency according to the frequency of the generated code.

[0003]

In the coding apparatus, a block having a large amount of motion in an image originally has a relatively lower resolution than a block having a small amount of motion, and a block having a small amount of motion has a larger amount of information. Need to be assigned. However, in the above-described conventional example, since a block having a large amount of motion and a block having a small amount of motion are quantized in the same quantization step, it is not possible to allocate a large amount of information to the block having a small amount of motion.

Therefore, according to the present invention, a moving block is quantized with a larger quantization step to suppress the amount of information generated, and a larger amount of information is assigned to a block having less motion, whereby the image quality is made uniform. It is an object of the present invention to provide an encoding device capable of performing the above.

[0005]

In the encoding device according to the present invention, classification is performed according to the amount of information of the image signal, and the classification step for switching the quantization step according to the value of the classification, and the image classification A motion detection means for detecting motion information is provided, and the values classified into the above classes are changed according to the motion detection value.

[0006]

According to the present invention, by changing the classification value to a value such that the quantization step becomes large when the motion is large, a larger amount of information is allocated to the image data having little motion. To be

[0007]

FIG. 1 is a block diagram showing an embodiment of the present invention. In FIG. 1, the image data input from the input terminal 1 is blocked by the blocker 2 into raster rows arranged in the order of scanning lines as shown in FIG. 2A, as shown in FIG. 2B. In this case, it is divided into blocks of 8 pixels horizontally and 8 pixels vertically. The block image data is converted to the orthogonal transformer 3
Is converted by. For example, as the orthogonal transformer 3, an orthogonal transform coding system called discrete cosine transform (hereinafter referred to as DCT) or the like is used.

At the same time, the image data blocked by the blocking unit 2 is also output to the motion detector 11. The motion detector 11 detects the motion of each block,
The motion detection value is output to the classifier 12.

The classifier 12 classifies the blocks to be processed into four classes according to the amount of information of the blocks (fineness of the image of the blocks), and each quantizer 5a,
It outputs to 5b, 5c and 5d, and also to the multiplexer 10. As a method of classifying the classifier 12, a method of classifying on the basis of the maximum absolute value of the AC component of the DCT conversion data, a method of classifying on the basis of the sum of squares of the AC component, or the like is used. In this embodiment, the quantization step is divided into four classes and the quantization step increases from 0 to 3.

In the quantizers 5a, 5b, 5c and 5d each having a different quantization step, the image data orthogonally transformed by the quantization step shown in FIG. 3 is quantized by the classification value. In other words, a block having a large amount of information is quantized more coarsely so that the amount of information generated in each block is made constant and the image quality is made uniform.

Further, in the classifier 12, since the human visual characteristics are deteriorated when the motion is large, the class classification of FIG. 3 is changed to a class classification value such that the quantization step becomes large as shown in FIG. Then, the coarser quantization is performed and the quantization is output to the quantizers 5a to 5d and the multiplexer 10. For example, when the class is usually 1, but it is determined that the motion is large, the class is set to 2 and coarser quantization than in the normal time is performed.

As described above, by changing the class value according to the amount of motion, the quantization step shown in FIG. 3 is roughly quantized by a large quantization step when the motion is large according to the class change of FIG. can do.

The quantized data is output to the selector 6 and also to the delay units 7a, 7b, 7c and 7d. The delay devices 7a to 7d delay the quantized data until the output result of the selector 6 is output, and then output the quantized data to the terminals 8a, 8b, 8c and 8d of the switch 8. The selector 6 switches the switch 8 so as to select a quantizer having a predetermined amount of information or less in a predetermined processing unit, for example, 30 blocks, and outputs the number of the selected quantizer to the multiplexer 10. Next, the variable-length encoder 9 allocates a short codeword to a code having a high occurrence frequency and a long codeword to a code having a low occurrence frequency according to the frequency of the generated code, and generates an information amount. Is output to the multiplexer 10.

The multiplexer 10 multiplexes the compressed image data, the classification value output from the classifier 12 and the quantizer number output from the selector 6, and outputs the output terminal 13 Output more.

[0015]

As described above, according to the present invention, the movement of each block is detected, and the value of the classification of each block is changed according to this movement information. The actual quantization step can be switched without the need to newly add. Also, by selecting the quantization step according to the amount of motion of the image,
The image quality can be made uniform between the block having a large motion and the block having a small motion, and the effect of improving the image quality can be obtained.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention.

FIG. 2 is a configuration diagram showing how data is divided into blocks.

FIG. 3 is a configuration diagram showing a relationship between classification and quantization steps.

FIG. 4 is a configuration diagram for explaining an operation of changing a classification.

[Explanation of symbols]

 3 Orthogonal transformers 5a, 5b, 5c, 5d Quantizer 6 Selector 8 Switch 9 Variable length encoder 11 Motion detector 12 Classifier

Claims (1)

[Claims] 1. A classifying unit for classifying according to the amount of information of an orthogonally transformed image signal and switching a quantization step according to a value of the classifying; and a motion detecting unit for detecting a motion of the image signal. And a changing means for changing the value of the classification by the classification means according to the output of the motion detecting means.