JPH0564016A - Picture encoder and picture decoder - Google Patents

Abstract

PURPOSE:To improve the quality of a picture by providing a means to detect the change of activity, and making the quantizing step of the block of the higher activity than a peripheral block further smaller. CONSTITUTION:On a picture encoder in which the step width of quantization is changed by a block unit, an activity detecting means 6 to detect the activity of each block unit, the means (HPF) 21 to detect the change of the activity between the blocks, and a quantization class deciding meand 24 to decide the class value of the quantization from the change of the activity are provided. Then, the quantizing step of the block of the higher activity than the peripheral block is made still smaller. Thus, the quantization of an edge part or an isolated picture is made fine by adaptive quantization, and a mosquito noise is reduced, and the picture quality can be improved. Besides, by variable-length-encoding the class information of the adaptive quantization, more adequate adaptive quantization can be realized.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a high-efficiency coding system for digitizing an image with a smaller amount of data in a recording, transmitting and displaying device for processing an image signal, and more particularly to an image using adaptive quantization. The present invention relates to an encoding device and a decoding device.

[0002]

2. Description of the Related Art In high-efficiency image coding, there is adaptive quantization in which the step size of quantization is changed in block units according to its activity (activity and degree of change).
A block with high activity is difficult to detect even if the quantization error is large, and a block with low activity is likely to detect the quantization error. Therefore, a block having high block activity has a wide (coarse) quantization step width, and conversely, a block having low activity has a narrow (fine) quantization step width. As a result, the quantization is adapted to the visual characteristics and unnecessary fine quantization is eliminated, so that the amount of generated data can be reduced. It should be noted that the processing block is generally 8 × 8 pixels or the like in accordance with the size in the encoding using the orthogonal transform.

Further, in such adaptive quantization, since the information (quantization step width) needs to be transmitted to the decoding side, if there are many kinds of changes due to the activity, the information will increase. Therefore, it is general that the number of types of change due to adaptive processing is set to about four types as a quantization class. This quantization class is a value in the intermediate stage where the quantization step is determined from the activity.

FIG. 4 is a block diagram showing a conventional image coding apparatus. In the figure, the image signal input from the image input terminal 1 is supplied to the orthogonal transformer 2. The orthogonal transformer 2 outputs the input signal to the DC for each block of 8 × 8 pixels.
The orthogonal transformation is performed by a method such as T (discrete cosine transformation). One of the transform coefficients output from the orthogonal transformer 2 is a DC coefficient indicating the average value of the block, and the other is an AC coefficient indicating the state of change, and each coefficient is supplied to the adaptive quantizer 3.

The adaptive quantizer 3 quantizes each coefficient with a step width set by a method described later and supplies it to the variable length encoder 4. Adaptive quantization changes the quantization method depending on the nature of the image, but here, the step size of the quantization is changed by the block activity of the orthogonal transform coefficient.

The variable-length encoder 4 variable-length-encodes the quantized coefficient and outputs it as data from the data output terminal 5 to the decoding device. Here, since the AC coefficients are concentrated in the vicinity of 0, the variable-length code of the AC coefficients has the shortest code length of 0, and the code length increases as the absolute value increases, thereby reducing the data amount. ing.

On the other hand, the transform coefficient which is the output signal of the orthogonal transformer 2 is also supplied to the activity detector 6.
The activity detector 6 calculates the sum of absolute values of the AC coefficients of each block, and supplies it to the quantization class determiner 7 as an activity value A for each block. The quantization class determiner 7 determines the quantization class value C based on the activity value A. There are four class values C from 0 to 3, and the class value C increases by 1 every time A is doubled. The larger the number of classes, the more desirable it is in terms of characteristics, but the number of classes cannot be increased because the information needs to be transmitted. The class information output from the quantization class determiner 7 is transmitted from the class information output terminal 8 to the decoding device side, and is also supplied to the adaptive quantizer 3 described above.

[0008]

In the encoding using the orthogonal transform, the quantization error spreads in the block when the inverse transform is performed in the decoding device. Then, in the edge portion of the image, the quantization error spreads to the periphery of the edge. This is called mosquito noise and causes visual deterioration in image quality. When adaptive quantization is performed, the edge part has high activity (activity), so the quantization becomes rough, quantization noise increases, and mosquito noise increases, which is also appropriate for visual characteristics. However, there is a problem in that the class information of the adaptive quantization increases when the adaptive quantization is performed, and the total generated code amount cannot always be reduced.

On the other hand, the present applicant has already proposed "Image Coding Device and Decoding Device" (Japanese Patent Application No. 2-262389). This is a class value of the quantization determined by the detected activity, which is pass-processed by an LPF (low-pass filter), and the activity is lowered by an adjacent low activity block corresponding to a flat part in the edge part, This is a method of reducing the amount of information to be transmitted by thinning out the class value, because the quantization does not become coarser than necessary in the adaptive quantization, the image quality deterioration at the edge part can be improved.

However, in transform coding, the mosquito noise at the edge portion is the main cause of image quality deterioration, and it is desired that the edge portion be further improved as compared with other portions. Further, there remains a problem that quantization error is conspicuous even in an image existing in isolation in front of a flat background and improvement is desired.

The present invention has been made by paying attention to the above points, and has a means for detecting a change in activity, and improves the image quality by making the quantization step finer in a block having a higher activity than the surrounding blocks. An object of the present invention is to provide an image coding apparatus and an image decoding apparatus capable of reducing the amount of data to be transmitted by performing variable length coding of adaptive quantization class information.

[0012]

In order to solve the above-mentioned problems, the present invention provides (1) activity detection for detecting activity of each block in an image coding apparatus which changes the quantization step size in block units. Means 6;
It has a means (HPF21) for detecting a change between the blocks of the activity and a quantization class judgment means 24 for judging a class value of the quantization from the change of the activity, and a quantum of a block having a large activity with respect to a peripheral block. An image encoding device characterized in that the encoding step is made finer,

(2) In the image coding apparatus that changes the quantization step size in block units, the quantization class determination means 24 for determining the quantization class value and the difference value between the blocks of the quantization class value are variable. Provided is an image encoding device characterized by having means 25 for long encoding,

(3) In a decoding device for an image coded by changing the quantization step size in block units, variable length for decoding quantization class information transmitted by variable length code into fixed length code (EN) An image decoding device characterized by comprising a decoding means (17) and a means (13) for carrying out an adaptive dequantization based on the quantization class information obtained by the variable length decoding means.

[0015]

[Operation] Find the change between each block of activity,
Therefore, by changing the quantization class, the activity becomes higher than that of the peripheral block in the image existing independently in the edge portion or the flat background, and thus the quantization becomes fine. Therefore, mosquito noise is reduced, and quantization suitable for visual characteristics can be performed. Since the class of adaptive quantization has a strong correlation between blocks, the amount of code can be reduced by variable length coding of the difference between blocks.

[0016]

1 is a block diagram showing an embodiment of an image coding apparatus of the present invention. The same components as those of the conventional example shown in FIG. 4 are designated by the same reference numerals. In FIG. 1, FIG.
The difference is the method of determining the quantization class, and the activity passes through the LPF (low pass filter) 21 and the HPF (high pass filter) 22 and then the quantization class determiner 24
Used in. The configurations and operations of the orthogonal transformer 2, the adaptive quantizer 3, the variable length encoder 4, and the activity detector 6 are the same as those in the conventional example. The activity value A of each block output from the activity detector 6 is LPF21 and HPF22.
Entered in. The LPF 21 and the HPF 22 are spatial filters, which replace the processing for the pixel value of the normal LPF with the one for the activity value A for each block.

FIG. 2 is a two-dimensional representation of tap coefficients for each block of LPF21 and HPF22. The median values corresponding to the blocks subject to adaptive quantization are both
It is 1/2 (8/16), but the peripheral area is 1/16 for LPF and -1 for HPF.
/ 16. The output M of the LPF21 is the averaged activity of the target block and its neighboring blocks, and HP
The output D'of F22 represents the degree of change.

The output value M of the LPF 21 is input to the divider 23 and the quantizer class determiner 24. The output D of the HPF 22 is input to the divider 23. In the divider 23, D ′ / M is obtained as a normalized change (difference) D and is input to the quantization class determiner 24. The quantization class C, which is the output of the quantization class determiner, is supplied to the adaptive quantizer 3 and the differential variable length encoder 25.

The quantizer class determiner 24 determines the quantizer class C from D and M with the characteristics shown in FIG. In the determination characteristic, the value of D is within ± 0.5, and the upper limit of the value of M is also limited. Class value C judged is 0 to 4
The quantization step Sq is 1. every time C increases by 1.
It is increased by 3 times. The adaptive quantizer 3 quantizes the coefficient by multiplying the quantization step Sq determined by the class value C by the control coefficient k. k is a coefficient for controlling the amount of data, and is determined from the outside in order to obtain the desired amount of data. Therefore, it can be said that Sq is a relative value of quantization.

As described above, the determination of the quantization class is made based on the average component M and the change component D of the activity, and its characteristic will be described. When D is constant, the class becomes smaller and the quantization becomes finer as M becomes smaller, as in the conventional case. On the other hand, when M is constant, as the value of D increases, the class decreases and the quantization becomes finer.
The smaller the value of, the higher the class and the coarser the quantization. When the value of M is positive, the activity of the block is higher than that of the surroundings. This corresponds to an image or an edge portion existing in isolation in front of a flat background. In such a portion, the quantization error is noticeable and it is appropriate to make the quantization fine.

The quantization class C thus obtained needs to be transmitted to the decoder. Differential variable length encoder 25
Then, the difference between C one block before and C of the current block is obtained, and the value is encoded by the variable length code as shown in FIG. An example of variable length coding is shown in FIG. It can be seen that the number of classes has increased from 4 to 5 as compared with the conventional example, but the code amount has decreased (2 bits per block are required in the conventional example without variable length coding).

Next, the image decoding apparatus will be described.
FIG. 5 is a block diagram showing an embodiment of the image decoding apparatus of the present invention. In the figure, the compressed data transmitted from the data output terminal 5 of the encoder of FIG.
It is supplied to the variable length decoder 12 via 11. Variable length decoder
12 converts the variable length code into a normal code and supplies it to the dequantizer 13.

On the other hand, the class information input signal transmitted from the class information output terminal 8 of the encoding apparatus of FIG. 1 is supplied to the variable length decoder 17 via the class information input terminal 14. The variable length code 17 converts the variable town coded class information into a fixed length and supplies it to the inverse quantizer 13. Inverse quantizer 13
Replaces the code supplied from the variable length decoder 12 with a representative value for quantization and supplies the code to the orthogonal inverse transformer 15. The step size of the representative value replacement in the inverse quantizer is determined by the class value C as in the adaptive quantizer 3 in the encoding device. The quadrature inverse transformer 15 converts the input signal to an inverse DCT
It is converted to obtain a reproduced image signal and is output through the image output terminal 16.

[0024]

The image coding apparatus and the decoding apparatus of the present invention have extremely excellent effects as described below. (1) It has a means to detect the edge part of the image by the change of activity, and the quantization step is made finer in the edge part, so that the quantization of the edge part and the isolated image becomes finer by the adaptive quantization, and the mosquito noise is reduced. It is reduced and the deterioration of image quality is improved.

(2) Also, by variable-length coding the adaptive quantization class information, the amount of data that must be transmitted can be reduced, and even if the number of classes is increased, there is no significant change in the classes. The amount of data does not increase so much and more appropriate adaptive quantization becomes possible.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of an image encoding device according to the present invention.

FIG. 2 is a diagram showing two-dimensionally each tap coefficient of LPF and HPF.

FIG. 3 is a characteristic diagram for determining a class value C from an average component M and a change component D of block activity.

FIG. 4 is a block diagram showing a conventional image encoding device.

FIG. 5 is a block diagram showing an embodiment of an image decoding apparatus of the present invention.

FIG. 6 shows an example of a differential variable length code of a quantization class.

FIG. 7 shows an example of differential variable length coding of a quantization class.

[Explanation of symbols]

1 ... Image input terminal, 2 ... Orthogonal transformer, 3, ... Adaptive quantizer, 4 ... Variable length encoder, 5 ... Data output terminal, 6 ... Activity detector, 7, 24 ... Quantization class determiner, 8
… Class information output terminal, 11… Data input terminal, 12, 17…
Variable length decoder, 13 ... Inverse quantizer, 14 ... Class information input terminal, 15 ... Orthogonal inverse transformer, 16 ... Image output terminal, 21 ... LP
F, 22 ... HPF, 23 ... Divider, 25 ... Differential variable length encoder.

Claims (3)

[Claims] 1. An image encoding apparatus that changes a quantization step size in block units, an activity detecting unit that detects an activity in each block unit, a unit that detects a change between the blocks of the activity, and an activity detecting unit. An image coding apparatus, comprising: a quantizing class judging means for judging a quantizing class value from a change, and finely quantizing a block having a large activity with respect to a peripheral block. 2. An image coding apparatus which changes a quantization step size in block units, wherein: a quantization class judging means for judging a quantization class value; and a variable length code for a difference value between blocks of the quantization class value. An image coding apparatus, comprising: 3. A variable length decoding device for decoding a quantization class information transmitted by a variable length code into a fixed length code in a decoding device for an image coded by changing the quantization step size in block units. An image decoding apparatus comprising: means, and means for performing adaptive dequantization based on the quantization class information obtained by the variable length decoding means.