JP2000350206A - High efficiency coder - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain stable image quality by providing a code quantity calculation section that uses 2nd priority different from 1st priority in addition to a selected quantizer and a code quantity calculation section using 1st priority calculated in response to a characteristic of a block to the high efficiency coder so as to realize improvement of the image quality thereby avoiding overflow. SOLUTION: A 1st priority calculation section 103 calculates the 1st priority in the unit of blocks from block information in a state of a received original and received block information after orthogonal transform. Second priority calculation sections 106, 108 change the received 1st priority into a specific priority to generate the 2nd priority. First coding sections 104, 105 use a quantizer decided by an offset given by the 1st priority to quantize the received orthogonal transform data to calculate a code quantity of the block. Second coding sections 107, 109 use a quantizer decided by an offset given by the 2nd priority to quantize the received orthogonal transform data to calculate a code quantity of the block.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a high-efficiency encoding apparatus used for encoding, recording, and transmitting an image signal with high efficiency. It is intended to provide a code amount control method for keeping the code amount below a certain level.

[0002]

2. Description of the Related Art In recent years, with the development of digital compression technology,
A digital VTR that digitally compresses and records a video signal has attracted attention.

Here, the conditions required for the VTR are: high image quality;・ Resistant to errors. -Special playback is possible and sufficient image quality is realized. -The hardware scale is small. It is mentioned. In order to satisfy these conditions, it is desirable to divide the entire image into specific small code amount control units and perform code amount control such that the code control units fall within a specific code amount.

[0004] A conventional high-efficiency coding apparatus will be described below with reference to Figs.

In FIG. 6, reference numeral 200 denotes an input unit, 201 denotes a segment configuration unit, 202 denotes an orthogonal transform operation unit, 203 denotes a priority calculation unit, 204 and 205 denote coding units, 206 denotes a quantizer determination unit, and 207 denotes a quantization unit. An encoding unit 208 is an output unit.

Next, the operation of the conventional high efficiency coding system will be described. In the conventional example, two types of quantizers are selectable for simple description, and are expressed as a quantizer 0 and a quantizer 1, respectively. In addition, quantizer 0 is set to quantization No. Is 0, the quantizer 1 is the quantizer N
o. May be expressed as one quantizer, but it is considered that there is no particular confusion. The larger the number of the quantizer, the smaller the quantization step, that is, when the same orthogonal transform data is quantized and variable-length coded, the generated code amount is largest in the quantizer 1 and conversely, the quantizer 0 is The smallest. Also, there are two types of priorities calculated for each block, which are expressed as priority 0 and priority 1, respectively. The smaller the priority number, the higher the priority, and the larger the priority number, the lower the priority. That is, when the priority is expressed as high, it means that the numerical value of the priority is small. A block of priority 0 has a higher priority than a block of priority 1, and a block of priority 1 has a lower priority than a block of priority 0. In a segment, a block having a lower priority is quantized by a quantizer having a smaller quantization step. Conversely, a block having a higher priority is quantized by a quantizer having a larger quantization step.

[0007] The image signal input from the input section 200 is input to the segment forming section 201. The segment forming unit 201 divides an input image into blocks each including 8 × 8 pixels, and collects 20 blocks to form a segment which is a code amount control unit. Each segment is input to the orthogonal transform operation unit 202 and the priority calculation unit 203.

[0008] The orthogonal transform operation unit 202 performs orthogonal transform on the input segment in block units and outputs orthogonal transform data. The orthogonal transformation data is stored in the priority calculation unit 20
3, input to the encoder 204, the encoder 205, the quantizer determiner 206, and the encoder 207.

The priority calculating section 203 calculates the priority in units of blocks from the input block information in the original picture state and the input block information after the orthogonal transformation. The priority is represented by two types of numbers 0 and 1, as described above. The priority of each block is determined by the coding units 204, 205,
It is input to the quantizer determination unit 206.

[0010] The encoding unit 204 quantizes the input orthogonally transformed data by a quantizer determined by (0 + offset given by priority), and further performs variable length encoding.
The code amount of the block is calculated. The offset is uniquely determined for each priority. In the conventional example, the offset of priority 0 is 1 and the offset of priority 1 is 0. For example, when the priority of a certain block is 0, the encoding unit 2
In step 04, the block is quantized by the quantizer 1 and further subjected to variable-length coding to calculate a code amount. The same processing is performed on the 20 blocks in the segment to calculate the total code amount for each segment. The total code amount calculated by the encoding unit 204 is input to the quantizer determining unit 206.

Another encoding unit 205 quantizes the input orthogonally transformed data by a quantizer determined by (1 + offset given by priority), further performs variable length encoding, and performs Calculate the code amount. For example, when the priority of a certain block is 0, the encoding unit 205
, The block is quantized by the quantizer 1 and further subjected to variable-length coding to calculate a code amount. The same processing is performed on the 20 blocks in the segment to calculate the total code amount for each segment. The total code amount calculated by the encoding unit 205 is input to the quantizer determining unit 206. If (1 + offset given by priority) exceeds 1, the selected quantizer is quantizer 1.

In the prior art, the priority offset is 1 for priority 0 and 0 for priority 1, but other offsets such as 0 for priority 0 and -1 for priority 1 are also conceivable. When this offset is used, the quantizer No.
Becomes -1, and the selectable quantization No. May be smaller than the minimum value of the quantizer No. selected in this case. Becomes 0.

[0013] The quantizer determining unit 206 determines a quantization amount corresponding to a maximum code amount not exceeding a specific target code amount out of the total code amount in the encoding unit 204 and the total code amount in the encoding unit 205. Let the quantizer be the final quantizer. The final quantizer is input to the encoding unit 207.

The coding unit 207 quantizes the input orthogonally transformed data by a quantizer determined by (final quantizer + offset given by priority), and further performs variable length coding, and Is output. The encoded data is input to the output unit 208. If (final quantizer + offset given by priority) exceeds 1, the selected quantizer is quantizer 1.

The output unit 208 outputs the input encoded data to a recording / transmission medium or the like.

The determination of the final quantizer will be described with reference to FIG. FIG. 7 shows the code amount calculated by each coding unit of a certain segment S0. Segment S
The priority of the 20 blocks in 0 is all 1. That is, the offset in this case is zero. 240 is the code amount calculated by the encoding unit 1, the value is 3000 bits, 241 is the code amount calculated by the encoding unit 2, and the value is 4000 bits. Reference numeral 242 denotes a target code amount of the segment, and its value is 3500 bits. In this case, the generated code amount of the encoding unit 1, which is the maximum code amount not exceeding the target code amount, becomes the final code amount, and the quantizer 0 in the encoding unit 1 becomes the final quantizer in the segment. .

[0017]

However, in the high-efficiency coding apparatus shown in the conventional example, when the priority of a block is high, even if coding is performed using a quantizer having the largest quantization step, There is a first problem that the generated code amount of the segment exceeds the target code amount. This phenomenon is called overflow. The overflow, which is the first problem, will be described with reference to FIG.

FIG. 8 shows the code amount calculated by each coding unit of a certain segment S0. Note that the priority of all 20 blocks in the segment S0 is 1. That is, the offset in this case is zero. 250 is the code amount calculated by the coding unit 1, the value is 4000 bits, 251 is the code amount calculated by the coding unit 2, the value is 5000 bits, 252 is the target code amount of the segment, Its value is 3500 bits. Encoding unit 1
, And the code amount in the encoding unit 2 both exceed the target code amount, and overflow has occurred.
In this case, the quantizer 0 in the encoding unit 1 is the final quantizer.

When an overflow occurs, since the generated code amount in the segment exceeds the code amount allocated to the segment, all data is not recorded and some data is discarded. Since the encoded data in the segment is recorded in the recording area of the recording medium in accordance with the priority order between the blocks set in the segment, the higher-priority block among the 20 blocks constituting the segment Is recorded in the recording area, whereas data of other blocks with lower priority is hardly recorded. Therefore, in the reproduced image, blocks in which all or most of the encoded data are recorded have clear image quality including high-frequency components, but blocks in which almost no encoded data are recorded have blurs in which the high-frequency components are lost. Resulting in a very high quality image.

There is also a technique for reducing the number of bits for a block having a specific priority for the purpose of reducing the amount of information. For example, it is conceivable to set the priority of a block having an AC coefficient exceeding a specific value to 1, and to divide all the AC coefficients by 2 before quantizing a block having a priority of 1.
In this case, the offset of the priority 1 is set to 0.

Although this is an effective means for reducing the amount of information,
For a block with a priority set to 1, all AC coefficients are always divided by 2 even if encoding is performed using a quantizer with the smallest quantization step. As a result, there is a second problem that a generated code amount of the segment is much smaller than the target code amount, and a code amount allocated to the segment cannot be used effectively. The second problem will be described with reference to FIG.

In FIG. 9, reference numeral 260 denotes a code amount calculated by the coding unit 1 and its value is 500 bits, 261 denotes a code amount calculated by the coding unit 2 and its value is 1000 bits. Reference numeral 262 denotes a target code amount of the segment, and its value is 3500 bits. In this case, the quantizer 1 in the encoder 2 is the final quantizer, but the encoder 2
Is 1500 bits and the target code amount is 350 bits.
Since there are 0 bits, there is a margin of 2000 bits in the segment. When calculating the code amount in the encoding unit 2, each block is encoded using the priority and the quantizer 2. For the block with the priority 1, all the AC coefficients are set to 2 before quantization. That is, before the quantization by the quantizer 2, the AC coefficient is quantized with the quantization step of 2. Even if encoding is performed using the quantizer 2, the priority is 1 even though the generated code amount is within the target code amount, so that all AC coefficients are divided by 2 to allocate to segments. The used code amount cannot be used effectively. In particular, when the priority is set to 1 in a test image or the like having a small amount of information, the generated code amount of the segment is overwhelmingly small with respect to the target code amount.
A small AC coefficient is lost because the C coefficient is divided by 2,
The reproduced image will be distorted.

The present invention has been made in order to solve the above-mentioned problems, and has been made in accordance with the L quantizers selected from the selectable K quantizers and the L-th quantizers calculated according to the characteristics of the blocks. By setting T code amount calculation units using a second priority different from the first priority in addition to the L code amount calculation units using one priority, allocation in a segment is performed. To effectively use the amount of code that can be used to improve image quality,
It is another object of the present invention to provide a high-efficiency coding apparatus that avoids overflow and realizes stable image quality.

[0024]

According to a first aspect of the present invention, an input image is divided into blocks, and M blocks (M is a positive integer) are collected to form a segment. Means, orthogonal transform means for performing orthogonal transform on this segment in block units to generate orthogonal transform data, and S (S is a positive integer) controlling a quantizer selected for each block in block units First priority calculating means for calculating a first priority of
L (L is L ≦ L) among the quantizers (K is a positive integer)
(A positive integer that satisfies K), and performs encoding for each of the L quantizers from the orthogonal transform data, the first priority, and the L quantizers, A first code amount calculating means for calculating the L first code amounts, and T (T is a positive integer) quantizers selected from the L quantizers.
For each of the quantizers, a second priority calculating means for changing a specific priority and newly setting a second priority among the S first priorities, and orthogonal transform data; A second code amount calculating unit that performs coding for each of the T quantizers from the second priority and the T quantizers, and calculates T second code amounts; A quantizer determining means for determining a final quantizer and a final priority of a block from the first code amount, the T second code amounts, and a specific target code amount; a final quantizer and a final priority; Further, the present invention provides a high-efficiency encoding apparatus characterized by further comprising encoding means for quantizing and variable-length encoding blocks in a segment to generate encoded data.

According to a second aspect of the present invention, the second code amount calculating means encodes the orthogonal transform data using a quantizer having the lowest priority and the largest quantization step. It is a high-efficiency encoding device described in the above.

According to a third aspect of the present invention, an input image is divided into blocks, M blocks are collected to form a sync block, and N sync blocks (N is a positive integer) are collected to form a segment. Constituent means, orthogonal transform means for performing orthogonal transform on this segment in block units to generate orthogonal transform data, and S first priority control means for controlling a quantizer selected for each block in block units First priority calculating means for calculating the degree, and for the block having the lowest first priority, the AC component of the orthogonal data of the block is divided by 2 before quantization to generate a segment. In a system having code amount control means for controlling the code amount to be equal to or less than a specific target code amount, the code amount control means sets L quantizers among the selectable K quantizers to L. Select and straight The conversion data, the first priority and L
First code amount calculating means for coding using L quantizers and calculating L first code amounts, and T out of L quantizers
Second quantizers that select a particular one of the S first priorities and set a new second priority for each of the T first quantizers. And a second code amount calculating means for encoding using orthogonal transform data, a second priority, and T quantizers, and calculating T second code amounts. , L first code amounts, T second code amounts, and a specific target code amount, a quantizer determining means for determining a final quantizer and a final priority of a block in a segment; A high-efficiency coding apparatus comprising: a coding unit that quantizes and variable-length codes blocks in a segment based on a finalizer and final priority to generate coded data.

In a fourth aspect of the present invention, the second code amount calculating means converts the orthogonal transform data into a quantizer having the lowest priority and the largest quantization step, and a quantizer having the highest priority and the smallest quantization step. 4. The high-efficiency encoding apparatus according to claim 3, wherein encoding is performed using an encoder.

According to a fifth aspect of the present invention, the quantizer determining means provides the maximum code amount not exceeding the target code amount among the L first code amounts and the T second code amounts. 4. The high-efficiency encoding apparatus according to claim 1, wherein the priority and the priority are respectively a final quantizer and a final priority.

[0029]

(Embodiment 1) Hereinafter, a first embodiment of the present invention will be described with reference to FIGS.

In FIG. 1, reference numeral 100 denotes an input unit, 101 denotes a segment configuration unit, 102 denotes an orthogonal transform operation unit, 103 denotes a first priority calculation unit, and 104 denotes first encoding units 1 and 10.
5 is a first encoder 2, 106 is a second priority calculator 1, 107 is a second encoder 1, 108 is a second priority calculator 2, and 109 is a second encoder. Reference numerals 2 and 110 are a quantizer determining unit, 111 is an encoding unit, and 112 is an output unit.

Next, the operation of this embodiment will be described. In the present embodiment, two types of quantizers are selectable for simplicity of description, and are represented as a quantizer 0 and a quantizer 1, respectively. Quantizer 0 is quantized N
o. Is 0 and the quantizer 1 is the quantizer No. Is 1
May be expressed as a quantizer. The larger the number of the quantizer, the smaller the quantization step, that is, when the same orthogonal transform data is quantized and variable-length coded, the generated code amount is largest in the quantizer 1 and conversely, the quantizer 0 is The smallest. Also, there are two types of priorities calculated for each block, which are expressed as priority 0 and priority 1, respectively. As described in the description of the conventional example, the smaller the priority number is, the higher the priority is, and the larger the priority number is, the lower the priority is. That is, when the priority is expressed as high, it means that the numerical value of the priority is small. A block of priority 0 has a higher priority than a block of priority 1, and a block of priority 1 has a lower priority than a block of priority 0. In a segment, a block having a lower priority is quantized by a quantizer having a smaller quantization step. Conversely, a block having a higher priority is quantized by a quantizer having a larger quantization step.

The image signal input from the input section 100 is input to the segment forming section 101. The segment forming unit 101 divides an input image into blocks each including 8 × 8 pixels, and collects 20 blocks to form a segment that is a code amount control unit. Each segment is input to the orthogonal transform operation unit 102 and the first priority calculation unit 103.

The orthogonal transform operation unit 102 performs an orthogonal transform on the input segment in block units and outputs orthogonal transform data. The orthogonal transform data is input to the first priority calculation unit 103, the first encoding units 104 and 105, the second encoding units 107 and 109, and the encoding unit 111.

The first priority calculation section 103 calculates a first priority in block units from the input block information in the original image state and the input block information after the orthogonal transformation. The priority is represented by two types of numbers 0 and 1, as described above. The first priority of each block is determined by a first encoding unit 104, 105, a second priority calculation unit 106,
08 and the quantizer determination unit 110.

The second priority calculating section 106 changes a specific priority with respect to the input first priority and generates a second priority. The second priority is the second encoding unit 107,
And input to the quantizer determination unit 110.

Another second priority calculating unit 108 includes:
A specific priority is changed from the input first priority to generate a second priority. The second priority is input to the second encoding unit 109 and the quantizer determination unit 110.

The first encoding unit 104 quantizes the input orthogonal transform data using a quantizer defined by (0 + offset given by the first priority), and further performs variable length encoding. The code amount of the block is calculated. The offset is uniquely determined for the priority. In the present embodiment, the offset of priority 0 is 1 and the offset of priority 1 is 0. For example, when the priority of a certain block is 0, the first coding unit 105 quantizes the block by the quantizer 1 and further performs variable length coding to calculate the code amount.
The same processing is performed on the 20 blocks in the segment to calculate the total code amount for each segment. The total code amount calculated by the first coding unit 104 is the quantizer determination unit 1
10 is input.

Another first encoding unit 105 quantizes the input orthogonal transform data by a quantizer determined by (1 + offset given by first priority), and further performs variable length encoding. To calculate the code amount of the block. For example, when the priority of a certain block is 0, the first coding unit 105 quantizes the block by the quantizer 1 and further performs variable length coding to calculate the code amount. The same process is performed on the 20 blocks in the segment.
Calculate the total code amount in segment units. The total code amount calculated by the first coding unit 105 is input to the quantizer determination unit 110. If (1 + offset given by priority) exceeds 1, the selected quantizer is quantizer 1.

The second priority calculation unit 106 changes a specific priority to the input first priority and generates a second priority. The second priority calculated by the second priority calculation unit 106 is input to the second encoding unit 107 and the quantizer determination unit 110.

Another second priority calculating unit 108 includes:
A specific priority is changed from the input first priority to a second priority. The second priority calculated by the second priority calculation unit 108 is equal to the second encoding unit 109,
And input to the quantizer determination unit 110.

The second encoding unit 107 quantizes the input orthogonal transform data using a quantizer determined by (0 + offset given by the second priority), and further performs variable length encoding. The code amount of the block is calculated. The same process is performed on the 20 blocks in the segment.
Calculate the total code amount in segment units. The total code amount calculated by the second coding unit 107 is input to the quantizer determination unit 110.

Another second encoding unit 109 quantizes the input orthogonal transform data by a quantizer determined by (1 + offset given by second priority), and further performs variable length encoding. To calculate the code amount of the block. The same processing is performed on the 20 blocks in the segment to calculate the total code amount for each segment. The total code amount calculated by the second encoding unit 109 is input to the quantizer determining unit 110. If (1 + offset given by priority) exceeds 1, the selected quantizer is quantizer 1. In this embodiment, the priority offset is set to 1 for priority 0 and 0 for priority 1. However, unexpectedly, offsets of 0 for priority 0 and -1 for priority 1 can be considered. When this offset is used, the quantizer No. Becomes -1, and the selectable quantization No. May be smaller than the minimum value of the quantizer No. selected in this case. Becomes 0.

The quantizer determining unit 110 receives the total code amount in the first encoding unit 104, the total code amount in another first encoding unit 105, and the second encoding unit 10.
7, another second encoding unit 109
, A final code amount is determined from the total code amount and the target code amount, and a quantizer and a priority corresponding to the final code amount are defined as a final quantizer and a final priority, respectively.
For example, if the total code amount in the first encoding unit 105 is the final code amount, the quantizer 1 is set to the final quantizer, the first priority is set to the final priority, and the second encoding unit is set to the final priority. If the total code amount in 107 is the final code amount, the quantizer 0 is the final quantizer, and the second priority in the second priority calculation unit is the final priority. The final quantizer and the final priority are input to the encoding unit 111.

The encoding unit 111 quantizes the input orthogonally transformed data by a quantizer determined by (final quantizer + offset given by final priority), and further performs variable length encoding. Output the encoded data of the block. The encoded data is input to the output unit 112. Note that (final quantizer + offset given by priority) is 1
Is exceeded, the selected quantizer becomes the quantizer 1.

The output unit 112 outputs the input coded data to a recording / transmission medium or the like.

Referring to FIG. 2, a certain segment S0
The determination of the final quantizer and the final priority will be described. In this segment S0, the total code amount in the first encoding unit all exceeds the target code amount.
In FIG. 2, reference numeral 150 denotes the total code amount in the encoding unit 104, the value of which is 4000 bits;
05 is the total code amount, the value is 5000 bits, 152 is the total code amount in the encoding unit 107,
The value is 3000 bits, 153 is the total code amount in the encoding unit 109, and the value is 6000 bits.
154 is a target code amount of the segment, and its value is 35
00 bits. In this case, the quantizer 0 corresponding to the encoding unit 107 becomes the final quantizer, and the second priority calculated by the second priority calculation unit 106 becomes the final priority.

Next, the determination of the final quantizer and the final priority in a certain segment S1 will be described with reference to FIG. In this segment S1, the total code amount in the encoding unit 104 falls within the target code amount, but the total code amount in the encoding unit 105 exceeds the target code amount. In FIG. 3, reference numeral 160 denotes a total code amount in the encoding unit 104, and its value is 2000 bits, 16 bits.
1 is the total code amount in the encoding unit 105, its value is 4000 bits, 162 is the total code amount in the encoding unit 107, its value is 1000 bits, and 163 is the total code amount in the encoding unit 109. , Its value is 3000 bits. 164 is a target code amount of the segment,
Its value is 3500 bits. In this case, the encoding unit 1
Quantizer 0 corresponding to 04 becomes the final quantizer, and the first priority becomes the final priority. The total code amount in the encoding unit 109 is 3000 bits, and under the condition that the maximum code amount does not exceed the target code amount, the encoding unit 109 is most suitable. Since the encoding unit 104 has a higher priority than the encoding unit, the encoding unit 104 has the final code amount.

As described above, the L quantizers are selected from the selectable K quantizers, and the orthogonal transform data of each block is converted to the first priority and the L quantizers. In addition to the L code amount calculators that perform coding using the code amount and calculate the code amount for each quantizer, a new code that calculates the code amount with a second priority different from the first priority By setting the amount calculation unit, it is possible to avoid overflow and provide stable image quality.

(Embodiment 2) Hereinafter, a second embodiment of the present invention will be described with reference to FIG.

The basic configuration and operation are as described in the first embodiment. In the present embodiment, the second priority calculating unit 106 sets all the second priorities to the lowest priority 1 and sets the second encoding unit 107
However, the encoding is performed by the quantizer 0 having the largest quantization step. In this case, encoding section 107
Is the minimum generated code amount.

As described above, L quantizers are selected from the selectable K quantizers, and the orthogonal transform data of each block is converted to the first priority and the L quantizers. In addition to the L code amount calculators that perform coding using the code amount and calculate the code amount for each quantizer, a new code that calculates the code amount with a second priority different from the first priority Set the quantity calculator. In this new code amount calculation unit, by giving a combination of a priority that minimizes the generated code amount and a quantizer,
Overflow can be avoided and stable image quality can be provided.

Embodiment 3 Hereinafter, a third embodiment of the present invention will be described with reference to FIG.

The basic configuration is as described in the first embodiment.

Next, the operation of this embodiment will be described. In the present embodiment, two types of quantizers are selectable for simplicity of description, and are represented as a quantizer 0 and a quantizer 1, respectively. Quantizer 0 is quantized N
o. Is 0 and the quantizer 1 is the quantizer No. Is 1
May be expressed as a quantizer. The larger the number of the quantizer, the smaller the quantization step, that is, when the same orthogonal transform data is quantized and variable-length coded, the generated code amount is largest in the quantizer 1 and conversely, the quantizer 0 is The smallest. Also, there are two types of priorities calculated for each block, which are expressed as priority 0 and priority 1, respectively. The smaller the priority number, the higher the priority, and the larger the priority number, the lower the priority. That is, when the priority is expressed as high, it means that the numerical value of the priority is small. Priority 0 blocks have higher priority than priority 1 blocks,
The priority 1 block has a lower priority than the priority 0 block. In a segment, a block having a lower priority is quantized by a quantizer having a smaller quantization step. Conversely, a block having a higher priority is quantized by a quantizer having a larger quantization step. Further, for a block having a priority of 1, all AC coefficients are reduced by 2 before quantization.
Divide by.

The image signal input from the input unit 100 is input to the segment forming unit 101. The segment forming unit 101 divides an input image into blocks each including 8 × 8 pixels, and collects 20 blocks to form a segment that is a code amount control unit. Each segment is input to the orthogonal transform operation unit 102 and the first priority calculation unit 103.

The orthogonal transform operation unit 102 performs orthogonal transform on the input segment in block units, and outputs orthogonal transform data. The orthogonal transform data is input to the first priority calculation unit 103, the first encoding units 104 and 105, the second encoding units 107 and 109, and the encoding unit 111.

The first priority calculation unit 103 calculates a first priority in block units from the input block information in the original image state and the input block information after the orthogonal transformation. The priority is represented by two types of numbers 0 and 1, as described above. The first priority of each block is determined by a first encoding unit 104, 105, a second priority calculation unit 106,
08 and the quantizer determination unit 110.

The second priority calculating section 106 changes a specific priority to the input first priority and generates a second priority. The second priority is the second encoding unit 107,
And input to the quantizer determination unit 110.

Another second priority calculation unit 108 includes:
A specific priority is changed from the input first priority to generate a second priority. The second priority is input to the second encoding unit 109 and the quantizer determination unit 110.

If the input orthogonal transform data has a priority of 1, the first encoder 104 divides all AC coefficients by 2 to obtain new orthogonal transform data. Subsequently, quantization and variable length coding are performed by a quantizer determined by (0 + offset given by the first priority), and the code amount of the block is calculated. The same processing is performed on the 20 blocks in the segment to calculate the total code amount for each segment. First encoding unit 104
Is input to the quantizer determining unit 110.

When the priority is 1 with respect to the input orthogonally transformed data, another first encoding unit 105 sets a value obtained by dividing all AC coefficients by 2 as new orthogonally transformed data. Subsequently, quantization and variable length coding are performed by a quantizer determined by (1 + offset given by the first priority), and the code amount of the block is calculated. The same process is performed on the 20 blocks in the segment.
Calculate the total code amount in segment units. Encoding unit 10
The total code amount calculated in 5 is input to the quantizer determining unit 110. (1 + offset given by priority) is 1
Is exceeded, the selected quantizer becomes the quantizer 1.

The second priority calculating unit 106 changes a specific priority with respect to the input first priority and generates a second priority. The second priority calculated by the second priority calculation unit 106 is input to the second encoding unit 107 and the quantizer determination unit 110.

Another second priority calculation unit 108
A specific priority is changed from the input first priority to generate a second priority. The second priority calculated by the second priority calculation unit 108 is the second priority of the second encoding unit 10
9 and the quantizer determination unit 110.

If the input orthogonal transform data has a priority of 1, the second encoding unit 107 divides all the AC coefficients by 2 to obtain new orthogonal transform data. Subsequently, quantization and variable length coding are performed by a quantizer determined by (0 + offset given by the second priority), and the code amount of the block is calculated. The same processing is performed on the 20 blocks in the segment to calculate the total code amount for each segment. Second encoding section 107
Is input to the quantizer determining unit 110.

When the priority is 1 with respect to the input orthogonally transformed data, another second encoding unit 109 sets a value obtained by dividing all AC coefficients by 2 as new orthogonally transformed data. Subsequently, quantization and variable length coding are performed by a quantizer determined by (1 + offset given by the second priority), and the code amount of the block is calculated. The same process is performed on the 20 blocks in the segment.
Calculate the total code amount in segment units. Encoding unit 10
The total code amount calculated in 9 is input to the quantizer determination unit 110. (1 + offset given by priority) is 1
Is exceeded, the selected quantizer becomes the quantizer 1. In this embodiment, the priority offset is set to 1 for priority 0 and 0 for priority 1, but unexpectedly, the priority offset is 0.
, And an offset of -1 with a priority of 1 can be considered.
When this offset is used, the quantizer No. Becomes -1, and the selectable quantization No. May be smaller than the minimum value of the quantizer No. selected in this case. Becomes 0.

The quantizer determining section 110 receives the total code amount in the first encoding section 104, the total code amount in another first encoding section 105, and the second encoding section 10.
7, another second encoding unit 109
, A final code amount is determined from the total code amount and the target code amount, and a quantizer and a priority corresponding to the final code amount are defined as a final quantizer and a final priority, respectively.
For example, if the total code amount in the first encoding unit 105 is the final code amount, the quantizer 1 is set to the final quantizer, the first priority is set to the final priority, and the second encoding unit is set to the final priority. If the total code amount in 107 is the final code amount, the quantizer 0 is the final quantizer, and the second priority in the second priority calculation unit is the final priority. The final quantizer and the final priority are input to the encoding unit 111.

If the input orthogonal transform data has a priority of 1, the encoding unit 111 divides all the AC coefficients by 2 as new orthogonal transform data. continue,
(Final quantizer + offset given by final priority)
Quantization and variable-length encoding are performed by the quantizer determined by, and encoded data of the block is output. The encoded data is input to the output unit 112. If (final quantizer + offset given by priority) exceeds 1, the selected quantizer is quantizer 1.

The output section 112 outputs the input coded data to a recording / transmission medium or the like.

Referring to FIG. 4, a certain segment S0
The determination of the final quantizer and the final priority will be described. In the segment S0, the total code amount in the first encoding unit is all within the target code amount. In FIG. 4, reference numeral 170 denotes the total code amount in the encoding unit 104, the value is 1000 bits, 171 is the total code amount in the encoding unit 105, the value is 1500 bits, and 172 is the total code amount in the encoding unit 107. And its value is 1800 bits, and 173 is
, And its value is 2000 bits. 174 is a target code amount of the segment, and its value is 3500 bits. In this case, the quantizer 1 corresponding to the encoding unit 109 becomes the final quantizer, and the second priority calculated by the second priority calculation unit 108 becomes the final priority.

As described above, L quantizers are selected from the selectable K quantizers, and the orthogonal transform data of each block is converted to the first priority and the L quantizers. In addition to the L code amount calculators that perform coding using the code amount and calculate the code amount for each quantizer, a new code that calculates the code amount with a second priority different from the first priority By setting the amount calculation unit, overflow and underflow can be avoided, and stable image quality can be provided.

(Embodiment 4) Hereinafter, a fourth embodiment of the present invention will be described with reference to FIGS.

The basic configuration and operation are as described in the third embodiment. In the present embodiment, the second priority calculating unit 106 sets all the second priorities to the lowest priority 1 and sets the second encoding unit 107
Perform encoding with the quantizer 0 having the largest quantization step, and further, the second priority calculation unit 108 sets all the second priorities to the highest priority 0, and It is characterized in that the quantization unit 109 performs encoding with the quantizer 1 having the smallest quantization step. In this case, the generated code amount in the encoding unit 107 becomes the minimum generated code amount, and the generated code amount in the encoding unit 109 becomes the maximum generated code amount.

Referring to FIG. 5, a certain segment S0
The determination of the final quantizer and the final priority will be described. In the segment S0, the total code amount in the first encoding unit is all within the target code amount. In FIG. 5, reference numeral 180 denotes a total code amount in the encoding unit 104, the value is 1000 bits, 181 is a total code amount in the encoding unit 105, its value is 1500 bits, and 182 is a total code amount in the encoding unit 107. And the value is 500 bits, 183 is the total code amount in the encoding unit 109, and the value is 3000 bits. 184 is a target code amount of the segment, and its value is 3500 bits. In this case, the quantizer 1 corresponding to the encoding unit 109 becomes the final quantizer, and the second priority calculated by the second priority calculation unit 108 becomes the final priority.

As described above, L quantizers are selected from the selectable K quantizers, and the orthogonal transform data of each block is converted to the first priority and the L quantizers. In addition to the L code amount calculators that perform coding using the code amount and calculate the code amount for each quantizer, a new code that calculates the code amount with a second priority different from the first priority Set the quantity calculator. At this time, in the new code amount calculation unit, the generated code amount is minimum,
By giving the maximum priority and the combination of the quantizer, overflow and underflow can be avoided and stable image quality can be provided.

(Embodiment 5) Hereinafter, a fifth embodiment of the present invention will be described with reference to FIGS.

The basic configuration and operation are as described in the first embodiment. In the present embodiment, quantizer determination section 110 determines the total code amount in encoding section 104, the total code amount in encoding section 105, the total code amount in encoding section 107, and the total code amount in encoding section 109. The maximum code amount that does not exceed the target code amount is set as the final code amount.

Referring to FIG. 3, a certain segment S0
The determination of the final quantizer and the final priority will be described. In the present embodiment, the generated code amount 163
Is the final code amount, the quantizer 1 corresponding to the second encoding unit 109 becomes the final quantizer, and the priority calculated by the second priority calculation unit 108 becomes the final priority.

As described above, L quantizers are selected from the selectable K quantizers, and the orthogonal transform data of each block is converted to the first priority and the L quantizers. In addition to the L code amount calculators that perform coding using the code amount and calculate the code amount for each quantizer, a new code that calculates the code amount with a second priority different from the first priority By setting the amount calculation unit, overflow and underflow can be avoided, and stable image quality can be provided. Furthermore, by setting the maximum code amount that does not exceed the target code amount as the minimum code amount in all of the first code amount and the second code amount, the code amount allocated to the code amount control unit can be effectively used. Can be used for

[0079]

As is clear from the above description, in the first invention of the present invention, in the code amount control for suppressing the generated code amount in a specific code amount control unit to be equal to or less than a predetermined value, the number of selectable K quantum L quantizers are selected from the quantizers, and the orthogonal transform data of each block is encoded using the first priority and L quantizers set based on the characteristics of the block. , A new T code amount calculator for calculating the code amount with a second priority different from the first priority in addition to the L code amount calculators for calculating the code amount for each quantizer By setting, overflow can be avoided and stable image quality can be provided.

In the second invention, in code amount control for suppressing the generated code amount in a specific code amount control unit to a certain value or less, L quantizers are selected from the selectable K quantizers. , The orthogonal transform data of each block is encoded by using the first priority and L quantizers, and is added to L code amount calculators that calculate the code amount for each quantizer. New T code amount calculation units for calculating the code amount with the second priority different from the one priority are set. In this new code amount calculation unit, by giving a combination of a priority and a quantizer that minimize the generated code amount, it is possible to avoid overflow and provide stable image quality.

According to the third aspect of the present invention, in the code amount control for dividing the total AC coefficient of the block having the lowest priority before quantization by 2 to suppress the generated code amount in a specific code amount control unit to a certain value or less. , Of the selectable K quantizers,
L quantizers are selected, and the orthogonal transform data of each block is coded using the first priority set based on the feature of the block and the L quantizers. In addition to the L code amount calculators that calculate the code amount, T new code amount calculators that calculate the code amount with a second priority different from the first priority are set. , Overflow can be avoided, and stable image quality can be provided.

In the fourth invention, in the code amount control for suppressing the generated code amount in a specific code amount control unit to a certain value or less by dividing all the AC coefficients by 2 for the block having the lowest priority before quantization. , Of the selectable K quantizers,
L quantizers are selected, and the orthogonal transform data of each block is coded using the first priority set based on the feature of the block and the L quantizers. In addition to the L code amount calculators for calculating the code amount, new T code amount calculators for calculating the code amount with a second priority different from the first priority are set. In this new code amount calculation unit, the priority that minimizes the generated code amount, the combination of the quantizer, and the priority that maximizes the generated code amount,
By providing a combination of an image and a quantizer, overflow and underflow can be avoided and stable image quality can be provided.

In the fifth invention, in code amount control for suppressing the generated code amount in a specific code amount control unit to a certain value or less, L quantizers are selected from the selectable K quantizers. , The orthogonal transform data of each block is coded using the first priority set based on the feature of the block and L quantizers, and the L number of quantizers are calculated for each quantizer. In addition to the code amount calculation unit, a new T code amount calculation units for calculating the code amount with a second priority different from the first priority are set, and the L code amount and the T code By setting the maximum code amount not exceeding the target code amount as the final code amount,
Overflow and underflow can be avoided and stable image quality can be provided.

[Brief description of the drawings]

FIG. 1 is a block diagram of first to fifth embodiments of the present invention.

FIG. 2 is a diagram showing a code amount of each coding unit for a first arbitrary segment according to the first embodiment of the present invention;

FIG. 3 is a diagram showing a code amount of each coding unit for a second arbitrary segment in the first and fifth embodiments of the present invention.

FIG. 4 is a diagram showing a code amount of each coding unit for a first arbitrary segment in a third embodiment of the present invention.

FIG. 5 is a diagram illustrating a code amount of each coding unit for a second arbitrary segment according to the fourth embodiment of the present invention.

FIG. 6 is a block diagram of a conventional example.

FIG. 7 is a diagram showing a code amount of each encoding unit for a first arbitrary segment in a conventional example.

FIG. 8 is a diagram showing a code amount of each coding unit for a second arbitrary segment in the conventional example.

FIG. 9 is a diagram showing a code amount of each coding unit for a third arbitrary segment in the conventional example.

[Explanation of symbols]

100, 200 Input unit 101, 201 Segment configuration unit 102, 202 Orthogonal transform operation unit 103 First priority calculation unit 104 First encoding unit 1 105 First encoding unit 2 106 Second priority calculation unit 1 107 Second encoder 1 108 Second priority calculator 2 109 Second encoder 2 110, 206 Quantizer determiner 111, 207 Encoder 112, 208 Output 150, 160, 170 , 180 Total code amount in the first encoder 1 151, 161, 171, 181 Total code amount in the first encoder 2 152, 162, 172, 182 Total code amount in the second encoder 1 153 , 163, 173, 183 Total code amount in second encoding section 2 154, 164, 174, 184, 242, 252, 252
62 Target code amount 204 Encoder 1 205 Encoder 2 240, 250, 260 Total code amount in encoder 1 241, 251, 261 Total code amount in encoder 2

 ────────────────────────────────────────────────── ─── Continuing on the front page (72) Inventor Mitsuhiro Miyashita 1006 Kazuma Kadoma, Kadoma-shi, Osaka Matsushita Electric Industrial Co., Ltd. F-term (reference) 5C053 FA02 FA21 GA11 GB05 GB22 GB26 GB32 KA04 KA21 KA24 LA06 5C059 KK29 KK35 LC01 MA21 MC11 MC30 MD02 ME01 PP04 SS11 SS12 TA46 TA52 TB08 TC15 TC19 TD14 UA02 UA05 5J064 AA02 BA16 BC02 BC16 BD02 BD03 5K041 AA00 BB00 CC04 CC07 EE31 FF31 FF36

Claims (21)

[Claims] 1. An input image is divided into blocks, and a segment forming means for forming a segment by collecting M blocks (M is a positive integer), and performing orthogonal transformation on the segment in block units to obtain an orthogonal Orthogonal transform means for generating transform data; and first priority for calculating S (S is a positive integer) first priority for controlling a quantizer selected for each block in the block unit. A degree calculating means, and L (L is a positive integer satisfying L ≦ K) quantizers are selected from among K selectable (K is a positive integer) quantizers, and the orthogonal transform is performed. Data, the first priority, and a first code amount calculating unit that performs encoding for each of the L quantizers from the L quantizers and calculates L first code amounts And selecting T (T is a positive integer) quantizers from the L quantizers For each of the T quantizers, a second priority calculating means for changing a specific priority among the S first priorities and newly setting a second priority, A second code amount for performing encoding for each of the T quantizers from the orthogonal transform data, the second priority, and the T quantizers, and calculating T second code amounts. Calculating means for determining a final quantizer in the segment and a final priority of the block from the L first code amounts, the T second code amounts, and a specific target code amount; Means, and coding means for quantizing and variable-length coding the blocks in the segment based on the final quantizer and the final priority to generate coded data. . 2. The method according to claim 1, wherein the second code amount calculating means encodes the orthogonal transform data using a quantizer having the lowest priority and the largest quantization step. High efficiency coding device. 3. A segment forming means for dividing an input image into blocks, collecting M blocks to form a sync block, and collecting N sync blocks (N is a positive integer) to form a segment; Orthogonal transform means for performing orthogonal transform on the segment in block units to generate orthogonal transform data; and S first priorities for controlling a quantizer selected for each block in the block unit. A first priority calculating means for calculating the first priority, for the block with the lowest first priority, divide the AC component of the orthogonal data of the block by 2 before performing quantization, In a system having code amount control means for controlling a generated code amount of a segment to be equal to or smaller than a specific target code amount, the code amount control means may select L out of K selectable quantizers. Select quantizer, the orthogonal transformation data,
First code amount calculating means for coding using the first priority and the L quantizers and calculating L first code amounts; T of the L quantizers; Of the S first priorities is changed for each of the T quantizers, and a second priority is newly set by changing a specific priority. And a second code amount calculation unit that encodes using the orthogonal transform data, the second priority, and the T quantizers to calculate T second code amounts. Means for determining a final quantizer and a final priority of the block in the segment from the L first code amounts, the T second code amounts, and a specific target code amount. Quantizer determining means; and quantizing, variable-length coding, and coded data of the block in the segment based on the final quantizer and final priority. And a coding means for generating data. 4. The quantizer according to claim 2, wherein the second code amount calculating means converts the orthogonal transform data into a quantizer having the lowest priority and the largest quantization step, and a quantizer having the highest priority and the smallest quantization step. 4. The high-efficiency encoding apparatus according to claim 3, wherein the encoding is performed by using. 5. The quantizer determining means for providing a maximum code amount not exceeding a target code amount among the L first code amounts and the T second code amounts, 4. The high-efficiency encoding apparatus according to claim 1, wherein the priority and the priority are a final quantizer and a final priority, respectively. 6. The second priority calculating means, for a specific quantizer among the T quantizers, generates an Rth (R is 2) for the first priority. ≤ R ≤ S) a low priority to the second lowest priority from the lowest priority to the lowest priority, and (R + 1) the lowest priority to the highest priority from the lowest priority to the lowest priority 4. The high-efficiency coding apparatus according to claim 1, wherein the priority is changed to one of (S-1) priorities excluding the priority. 7. The method according to claim 1, wherein the second priority calculating means determines a specific one of the T quantizers from an R-th highest priority with respect to the first priority. The highest priority is changed to the second highest priority, and any one of the (S-1) priorities excluding the highest priority is changed from the (R + 1) highest priority to the lowest priority. 4. The high-efficiency encoding device according to claim 1, wherein the encoding device is changed to a second one. 8. S = 4, R = 2, and the second priority calculating means performs the first priority calculation for a specific one of the T quantizers. The high efficiency coding apparatus according to claim 6, wherein the highest priority and the second highest priority are not changed. 9. S = 4, R = 2, and the second priority calculation means for the specific one of the T quantizers, The high efficiency coding apparatus according to claim 6, wherein the highest priority is not changed, and the second lowest priority is changed to the third highest priority. 10. S = 4, R = 2, and the second priority calculating means performs the first priority calculation for a specific one of the T quantizers. 7. The high-efficiency encoding apparatus according to claim 6, wherein the highest priority is changed to the second highest priority, and the second lowest priority is changed to the third highest priority. 11. S = 4, R = 3, and the second priority calculating means performs the first priority calculation for a specific one of the T quantizers. 7. The high efficiency coding apparatus according to claim 6, wherein the highest priority is not changed. 12. S = 4, R = 3, and the second priority calculation means performs the first priority calculation for a specific one of the T quantizers. 7. The high-efficiency coding apparatus according to claim 6, wherein the highest priority is changed to the second highest priority. 13. S = 4, R = 3, and the second priority calculating means performs the first priority calculation for a specific one of the T quantizers. 7. The high-efficiency encoding apparatus according to claim 6, wherein the highest priority is changed to the third highest priority. 14. S = 4, R = 2, and the second priority calculation means performs the first priority calculation for a specific one of the T quantizers. 8. The high-efficiency encoding apparatus according to claim 7, wherein the lowest priority and the second lowest priority are not changed. 15. S = 4, R = 2, and the second priority calculating means performs the first priority calculation for a specific one of the T quantizers. The high efficiency encoding apparatus according to claim 7, wherein the second lowest priority is not changed, and the lowest priority is changed to the second lowest priority. 16. S = 4, R = 2, and the second priority calculation means performs the first priority calculation for a specific one of the T quantizers. The high-efficiency encoding apparatus according to claim 7, wherein the lowest priority is changed to the second lowest priority, and the second lowest priority is changed to the third lowest priority. 17. S = 4, R = 3, and the second priority calculation means performs the first priority calculation for a specific one of the T quantizers. The high-efficiency encoding apparatus according to claim 7, wherein the lowest priority is not changed. 18. S = 4, R = 3, and the second priority calculation means performs the first priority calculation for a specific one of the T quantizers. 8. The high-efficiency encoding apparatus according to claim 7, wherein the lowest priority is changed to the second lowest priority. 19. S = 4, R = 3, and the second priority calculation means performs the first priority calculation for a specific one of the T quantizers. 8. The high-efficiency encoding apparatus according to claim 7, wherein the lowest priority is changed to the third lowest priority. 20. For a specific quantizer out of the T quantizers, the second priority calculation means sets all priorities to the lowest priority with respect to the first priority. 4. The high-efficiency encoding apparatus according to claim 1, wherein the priority is changed to a priority. 21. The second priority calculating means, for a particular quantizer among the T quantizers, sets all priorities to the highest priority with respect to the first priority. 4. The high-efficiency encoding apparatus according to claim 1, wherein the priority is changed to a priority.