JPH06268988A - Quantization characteristic control method for picture signal and picture signal compression-encoding device - Google Patents

Abstract

PURPOSE:To perform reproduction without damaging the decoded pictures even when signals for indicating the quantization bit number of picture signals are erroneous. CONSTITUTION:The picture signals 21 are orthogonally transformed in a blocking and orthogonal transformation part 1 and the quantization bit numbers are set for respective blocks in a bit allocation map setting part 2. A basic bit allocation map generation part 4 outputs a basic bit allocation map 24 and a basic quantization characteristic 25 is selected in a basic quantization characteristic selection part 5. A bit difference between a quantization bit allocation map 23 and the basic bit allocation map 24 is calculated in a bit difference calculation part 3 and the basic quantization characteristic is corrected in a quantization step width changing part 6. A quantization processing is performed in a quantization part 7 and a level in the basic quantization characteristic and the level of a corrected part are respectively separated into a basic code 28 and an additional code 29 and outputted.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a quantization characteristic control method and a compression coding apparatus for compressing and coding an image signal using orthogonal transformation.

[0002]

2. Description of the Related Art Conventional quantization characteristic control and compression encoding of image signals will be described with reference to the drawings.

FIG. 5 is a block diagram showing an example of an image signal compression coding apparatus using orthogonal transformation.

In FIG. 5, the input image signal 30 is subjected to blocking and orthogonal transformation processing in a blocking and orthogonal transformation unit 8. Orthogonal transform coefficient 3 obtained here
The bit allocation setting unit 9 adaptively allocates bits to 1, and a quantized bit number signal 32 representing the quantized bit number of each orthogonal transform coefficient is obtained. Then, the quantization characteristic corresponding to the number of quantization bits indicated by the quantization bit number signal 32 is obtained by the quantization characteristic selection unit 10, and the quantization characteristic 33 of each orthogonal transform coefficient is determined. Quantizer 1
In 1, the orthogonal transform coefficient 31 is quantized according to the quantization characteristic 33, and the quantized code 34 is output. On the other hand, the quantized bit number signal 32 is also output together with the code 34, and is used at the time of inverse quantization at the time of code.

[0005]

The quantized code 34 and the quantized bit number signal 32 are output from the compression encoding apparatus for image signals to the transmission path, and transmission, recording / reproduction, etc. are performed. At this time, if the quantized bit number signal 32 is erroneous on the transmission line, it is impossible to know how many bits the quantization process was performed in the inverse quantization process at the time of decoding. Therefore, there is a problem in that the correct decoding process is not performed and the decoded image fails.

An object of the present invention is to provide a quantization characteristic at the time of compression encoding in which a decoded image can be reproduced without breaking even if a signal representing the number of quantization bits of an orthogonal transform coefficient is erroneous in a transmission line. It is to provide a control method and a compression encoding device.

[0007]

A method for controlling a quantization characteristic of an image signal according to the present invention is to perform an orthogonal transformation on an input image signal for each block of a plurality of small areas having a predetermined size. With respect to the obtained orthogonal transform coefficient, a quantized bit allocation map indicating the first number of quantized bits of each orthogonal transform coefficient in the screen is set for each block, and the quantized bit allocation map In setting the quantization characteristic corresponding to the number of quantization bits of 1, the corresponding basic quantum from the basic bit allocation map indicating the second number of quantization bits predetermined for roughly quantizing the orthogonal transform coefficient. Of the second quantization bit number shown in the basic bit allocation map and the first quantization bit number shown in the quantization bit allocation map. The Tsu bets difference calculated to obtain the corrected quantization characteristic by correcting the quantization step width of the base quantization characteristics in accordance with the number of quantization levels represented by said bit difference.

The image signal compression coding apparatus according to the present invention comprises a block forming and orthogonal transforming unit which divides an input image signal into a plurality of small areas having a predetermined size and performs orthogonal transform on each block. A bit allocation map setting unit that sets a first quantization bit number for each of the blocks when quantizing each of the orthogonal transform coefficients obtained by the blocking and orthogonal transform units; A basic bit allocation map generator that outputs a basic bit allocation map indicating the bit allocation of each orthogonal transform coefficient in a block, and a basic bit allocation map generator that selects from among a plurality of predetermined quantization characteristics. A basic quantization characteristic selection unit for selecting a quantization characteristic corresponding to the second number of quantization bits of the orthogonal transform coefficient; A bit difference calculation unit for calculating a bit difference in the same coordinate among the block and the number of the second quantization bits indicated by the basic bit allocation map generation unit,
A quantization step width changing unit that corrects the quantization characteristic selected by the basic quantization characteristic selecting unit with the bit difference, and a quantum of the orthogonal transform coefficient according to the quantization characteristic corrected by the quantization step width changing unit. And a quantization unit that performs a quantization process.

[0009]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, the present invention will be described with reference to the drawings.

Referring to FIG. 1 showing an embodiment of the present invention, an image signal compression coding apparatus is provided with an input image signal 2
1 is divided into a plurality of small areas having a predetermined size and orthogonal transformation is performed for each block, and the orthogonal transformation coefficient 22 obtained by the blocking and orthogonal transformation unit 1 A bit allocation map setting unit 2 for setting the number of quantization bits 23 for each block for each block, and a basic bit allocation map showing the bit allocation of each orthogonal transform coefficient in a predetermined block. And a quantization characteristic corresponding to the quantization bit number 24 of each orthogonal transform coefficient indicated by the basic bit allocation map generation section 4 from among a plurality of predetermined quantization characteristics. The basic quantization characteristic selection unit 5 to be selected, the quantization bit number 23, and the quantization bit number 24 indicated by the basic bit allocation map generation unit 4 are blocked. , A bit difference calculation unit 3 that calculates a bit difference between the same coordinates, a quantization step width changing unit 6 that corrects the quantization characteristic selected by the basic quantization characteristic selection unit 5 with the bit difference, and this quantization step The quantization unit 7 performs a quantization process of the orthogonal transform coefficient 22 according to the quantization characteristic corrected by the width changing unit 6.

Next, the operation of the present embodiment will be described with reference to FIG. 3 showing a basic quantization characteristic and FIG. 4 showing a corrected quantization characteristic with reference to FIG. 1 and FIG. 2 showing a block configuration.

The input image signal 21 is subjected to blocking and orthogonal transformation in the blocking and orthogonal transformation unit 1. As shown in FIG. 2, the block formation is to divide the screen into a plurality of small area blocks, and the size of the block is generally about 8 pixels × 8 lines. One block has 64 pixels with a size of 8 lines. In image compression, for example, discrete cosine transform is generally selected as the orthogonal transform. Using the orthogonal transformation coefficient 22 subjected to orthogonal transformation, the bit allocation map setting unit 2 sets the number of quantization bits for quantizing each orthogonal transformation coefficient for each block.

Here, as shown in FIG. 2, there are 64 orthogonal transform coefficients per block, and each orthogonal transform coefficient A
_ij (i = 1,2,3 ... 8, j = 1,2,3 ... 8) is set as 64 quantization bit numbers.
The quantized bit number sequences are collectively referred to as a quantized bit allocation map.

Further, the orthogonal transform coefficient (A _ij ) represents the magnitude of the spatial frequency component which differs depending on the coordinates within the block, and the orthogonal transform coefficient A ₁₁ located at the upper left of the block represents the magnitude of the DC component within the block. And the magnitude of the high frequency component is shown toward the lower right. Therefore, in the quantized bit allocation map, since the DC component and the low frequency component have large values due to the nature of the image signal, a large number of quantization bits are allocated. The number of bits is often allocated in a small amount.

Further, in order to improve the compression efficiency, the quantized bit allocation map of each block is often different, and the quantized bit allocation map of a block with a fine pattern is allocated many bits up to high frequency components. However,
The quantized bit allocation map of the flat pattern block is such that a small number of bits are allocated from the low-frequency component. As described above, the quantized bit allocation map is set for each block and is output as the quantized bit number signal 23.

On the other hand, the basic bit allocation map generator 4 outputs a basic bit allocation map 24 showing the bit allocation of each orthogonal transform coefficient in a predetermined block. The basic bit allocation map 24 is for relatively coarsely quantizing each orthogonal transform coefficient in a block,
Commonly used for all blocks in the screen.
In this basic bit allocation map 24, a coarse quantization bit number is set such that the contents of the original image can be recognized even if the decoding process is performed after the quantization process is performed only in this map.

The basic quantization characteristic selection unit 5 selects a quantization characteristic corresponding to the number of quantization bits of each orthogonal transform coefficient indicated by the basic bit allocation map 24 from a plurality of predetermined quantization characteristics. For example, the characteristic 41 of FIG.
The quantization characteristic as shown in is selected. In this way, the quantization characteristic selected according to the basic bit allocation map 24 becomes the basic quantization characteristic 25.

Since the basic bit allocation map 24 is preset, it often does not match the quantized bit allocation map 23 created from the actual image signal. Further, since the number of quantization bits indicated by the basic bit allocation map 24 quantizes each orthogonal transform coefficient in the block relatively roughly, the quantized bit allocation map 23
It is often less than the number of quantization bits indicated by. This difference is not constant such that it is large in the quantized bit allocation map in the fine pattern block and small in the quantized bit allocation map in the flat pattern block.

Therefore, the difference is calculated by the bit difference calculation unit 3. The bit difference calculation unit 3 calculates the bit difference between the quantized bit number indicated by the quantized bit allocation map 23 and the quantized bit number indicated by the basic bit allocation map 24 at the same coordinates in the block.

For example, if the number of quantized bits indicated by the quantized bit allocation map is 5 and the number of quantized bits indicated by the basic bit allocation map is 3, the difference is 2 bits. If the number of quantized bits shown in the basic bit allocation map is larger than or equal to the number of quantized bits shown in the quantized bit allocation map, the difference becomes zero. The difference thus obtained is
It is output as the bit difference value 26.

Since the previously selected basic quantization characteristic 25 is selected based on the basic bit allocation map 24, it has a quantization characteristic corresponding to the fineness and flatness of the actual image. Absent. Therefore, the quantization characteristic is corrected using the bit difference value 26. The basic quantization characteristic 25 and the bit difference value 26 are input to the quantization step width changing unit 6. Then, the level value is obtained by calculating a power of 2 from the bit difference value 26. In the above example, since the bit difference value is 2, a level value of 4 is obtained.

That is, if the step width of the basic quantization characteristic is reduced to 1/4, that is, the level value is increased by 4, the quantization characteristic corresponding to the number of quantization bits indicated by the quantization bit allocation map can be obtained. it can.

The quantization characteristic correction will be described in detail below.
In FIG. 4, a characteristic 41 indicated by a broken line is a basic quantization characteristic. When the correction of the level value 4 of the previous example is added to this, the characteristic 42 shown by the solid line is obtained. Here, each level of the basic quantization characteristic 41 is simply divided into four to form a new level, but the correction method is not limited to this. The corrected quantization characteristic 42 has a level value four times that of the basic quantization characteristic 41, which means that the correction is performed using the bit difference value 26.

The quantization characteristic 27 corrected by the quantization step width changing unit 6 is used as the quantization characteristic when quantizing each orthogonal transform coefficient. In the quantizer 7,
Quantization processing is performed on the orthogonal transform coefficient 22 according to the corrected quantization characteristic 27. And output the obtained code,
At this time, the level of the basic quantization characteristic and the level of the corrected part are separately output to the basic code 28 and the additional code 29.

In the above example, the quantization bit number shown in the quantization bit allocation map 23 is 5 bits, but the quantization bit number shown in the basic bit allocation map 24 is 3 bits. For the orthogonal transform coefficient 22 input to the unit 7, first, the basic bit allocation map 2
The corresponding level on the basic quantization characteristic 25 obtained from 4 is represented by 3 bits, and the corresponding level of the level increased by the next correction is represented by 2 bits,
Each is output as a basic code 28 and an additional code 29.

This should be set so as to coincide with the index (3 bits) of the basic quantization characteristic when only the upper 3 bits of the index (5 bits) of each level value of the corrected quantization characteristic 27 are viewed. Then, it is possible to easily perform division output from the obtained 5-bit code such as the 3-bit basic code of the quantization bit number indicated by the higher-order basic bit allocation map and the remaining 2 bits as the additional code.

The basic code 28, the additional code 29, and the quantized bit number signal 23 thus obtained are output to the transmission line. On the decoding side, the inverse quantization process is performed with the quantization bit number according to the quantization bit number signal 23. Here, if the quantized bit number signal 23 is erroneous on the transmission line, the additional code 29 is not used, and only the basic code 28 is decoded according to the basic bit allocation map set in advance on the decoding side. By doing so, it is possible to reproduce the image using only the basic code 28, so that it is possible to prevent the image from collapsing and obtain an image of an image quality that allows the contents to be grasped.

Although the case where the signal indicating the number of quantized bits is directly output is described here, the same applies to the case where the number of quantized bit allocation maps is limited and the index of the map is output. It is possible.

[0029]

As described above, according to the present invention, an orthogonal transformation coefficient obtained by subjecting an input image signal to orthogonal transformation for each block of a plurality of small areas having a predetermined size is obtained. On the other hand, the first of the orthogonal transformation coefficients in the screen
When setting the quantization bit allocation map indicating the number of quantization bits of each block for each block and setting the quantization characteristic corresponding to the first quantization bit number indicated by the quantization bit allocation map, the orthogonal transform coefficient The corresponding basic quantization characteristic is selected from the basic bit allocation map indicating the second number of quantization bits predetermined for coarse quantization, and the second quantization bit number and the quantum indicated by the basic bit allocation map are selected. It is corrected by calculating a bit difference from the first quantization bit number indicated by the quantization bit allocation map and correcting the quantization step width of the basic quantization characteristic according to the quantization level number represented by the bit difference. By obtaining the quantized characteristics, when bits are adaptively assigned to each block when image compression coding is performed using orthogonal transform, the bits are Even if the additional information indicating the reliance is erroneous, decoding can be performed without corrupting the image, and the number of quantization bits can be set freely, so that it is possible to perform quantization characteristic control that hardly affects compression efficiency. it can.

[Brief description of drawings]

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

FIG. 2 is a diagram showing an example of the configuration of blocks in the present embodiment.

FIG. 3 is an explanatory diagram of basic quantization characteristics.

FIG. 4 is an explanatory diagram of quantization characteristic correction.

FIG. 5 is a block diagram showing a conventional example.

[Explanation of symbols]

1 block formation and orthogonal transformation unit 2 bit allocation map setting unit 3 bit difference calculation unit 4 basic bit allocation map generation unit 5 basic quantization characteristic selection unit 6 quantization step width changing unit 7 quantization unit 21 image signal 22 orthogonal transformation coefficient 23 Quantization Bit Allocation Map (Quantization Bit Number Signal) 24 Basic Bit Allocation Map 25 Basic Quantization Characteristic 26 Bit Difference Value 27 Corrected Quantization Characteristic 28 Basic Code 29 Additional Code 41 Basic Quantization Characteristic 42 Corrected Quantum Characteristics A ₁₁ , ..., A _ij , ..., A ₈₈ Orthogonal transform coefficient

Claims (2)

[Claims] 1. An orthogonal transform coefficient obtained by subjecting an input image signal to an orthogonal transform for each of a plurality of small-area blocks having a predetermined size In setting a quantized bit allocation map indicating the first number of quantized bits for each of the blocks and setting a quantization characteristic corresponding to the first quantized bit number indicated by the quantized bit allocation map, In order to roughly quantize the orthogonal transform coefficient, a corresponding basic quantization characteristic is selected from a basic bit allocation map indicating a predetermined second number of quantization bits, and the second basic bit allocation map shown in the basic bit allocation map is selected. A bit difference between the number of quantized bits and the first number of quantized bits indicated by the quantized bit allocation map, and a quantization level represented by the bit difference. A method for controlling a quantization characteristic of an image signal, wherein a corrected quantization characteristic is obtained by correcting a quantization step width of the basic quantization characteristic according to a number. 2. A block forming / orthogonal converting unit for dividing an input image signal into a plurality of small areas having a predetermined size and orthogonally converting each block, and the block forming / orthogonal converting unit. A bit allocation map setting unit that sets a first quantization bit number when quantizing each of the obtained orthogonal transform coefficients, and an orthogonal transform coefficient of each of the orthogonal transform coefficients in a predetermined block. A basic bit allocation map generator that outputs a basic bit allocation map indicating bit allocation, and a second quantization of each orthogonal transform coefficient indicated by the basic bit allocation map generator from a plurality of predetermined quantization characteristics. A basic quantization characteristic selecting unit for selecting a quantization characteristic corresponding to the number of bits, a first quantization bit number and the basic bit allocation map generating unit are shown. A bit difference calculation unit that calculates a bit difference between the second quantization bit number and the same coordinates in the block, and a quantization characteristic selected by the basic quantization characteristic selection unit with the bit difference. Image signal compression coding, which comprises a quantization step width changing unit for performing the quantization processing of the orthogonal transform coefficient according to the quantization characteristic corrected by the quantization step width changing unit. apparatus.