JP2000217114A - Apparatus and method for compression coding amount control - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an image compression coding method and an image compression coder, by which data can be most efficiently coded within a limited code amount. SOLUTION: This compression code amount control apparatus is provided with a block processing means 100, an orthogonal control means 101, a quantizing table 104, a quantizing means 102, a coding means 103, a code amount monitor means 106 and a stock storage means. Then the code amount monitor means 106 compares the object code amount per one block with a code amount of an output of the coding means 103 and controls the coding means 103 to conduct coding by adding part of the stocked code amount in the stock storage means to the object code amount, when the code amount of the output exceeds the object code amount. When the code amount of the stock is more than the exceeds code amount, the code amount used for the coding is subtracted from the stock and the means 106 forcibly terminates the coding of the block which is in progress at present, when the stocked code amount is less than the exceeded code amount.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a compression code amount control device and a compression code amount control method, and more particularly to a compression code amount control device and a compression code amount control method used for still images.

[0002]

2. Description of the Related Art As an example of a conventional compression encoding method, J.
The PEG compression encoding method will be described as an example. As shown in FIG. 4, a blocking unit 200 that divides image data into a plurality of blocks so as to have the same size, for example, a size of 8 × 8 pixels, and orthogonally transforms the blocked image data.
For example, orthogonal transform means 201 for performing DCT (discrete cosine transfer) transform, and transform coefficients output from the orthogonal transform
4 and a coding means 203 for compressing data output from the quantization means by a redundancy compression coding method such as Huffman coding. Conventionally, preprocessing has been performed to create data of the quantization table 204. In this preprocessing, first, compression is performed using a reference quantization table. That is, the image data is divided into
Is divided into a plurality of blocks each having a size of 8 × 8 pixels, and the block-formed image data is subjected to discrete cosine transform by the orthogonal transform means 201, and the transform coefficients are divided by the quantizing means by the coefficients indicated in the reference quantization table. The data output from the quantization means is compressed by a coding means using a redundancy compression coding method such as Huffman coding.

After performing this series of operations, the compressed code amount is checked. If this code amount is not within the range of the target code amount, a coefficient of the quantization table is calculated from the code amount of the compressed data, and this process is repeated several times to create compressed data having a desired code amount. I do. The compression operation is performed as follows using the quantized data determined by the above preprocessing. First, similarly to the above preprocessing, the image data is divided into a plurality of blocks having a size of 8 × 8 pixels by the blocking means 200, and the block-formed image data is subjected to discrete cosine transform by the orthogonal transform means 201, and the transform coefficients are quantized. The data is divided by the coefficient shown in the quantization table 204 by the means, and the data output from the quantization means is compressed by the coding means by a redundancy compression coding method such as Huffman coding. Here, the encoded generated code amount from the encoding unit 205 is monitored, and when the allocation to the block being encoded is exceeded, the processing of terminating the encoding of the block is performed.

[0004]

In such a conventional image compression encoding method and image compression encoding apparatus, in the pre-processing, one block is calculated from the ratio of the non-zero coefficient after quantization using the reference quantization table. It only assigns a code amount per unit, and does not consider data redundancy when encoding. Therefore, there is a problem that flexibility with respect to the data amount is small.

For example, even when the code amount per block after compression is smaller than the target value, only the remaining code amount is wasted. On the other hand, as described above, when the code amount exceeds the allocation amount, the coding is simply stopped, and no measures against data redundancy have been considered.

An object of the present invention is to provide an image compression encoding method and an image compression encoding apparatus which can encode the most efficiently within a limited code amount.

[0007]

According to the present invention, there is provided a compression code amount control apparatus, comprising: blocking means for dividing tangent image data into a plurality of blocks of a predetermined pixel size; and discrete cosine transform of the blocked image data. Orthogonal transform means, a quantization table for storing coefficients, a quantizing means for dividing a transform coefficient from the orthogonal transform means by a coefficient of the quantization table, and a code for compressing data output from the quantizing means. Encoding means, code amount monitoring means for monitoring the code amount per block of the encoding means, and means for storing a stock of a preliminary code amount calculated from a total target code amount of the image data, The code amount monitoring means compares the target code amount per block with the code amount of the output of the encoding means, and when the target code amount is exceeded, the code amount of the stock is increased. If the amount of code is equal to or more than the amount of code, the code amount used in the storage means is added to a part of the stored code amount in the storage means, control is performed using coding, and the used code amount is subtracted from the stock. When the amount of code stored is equal to or less than the amount of code exceeding the above, the coding of the block currently being coded is forcibly terminated.

[0008] Further, the code amount monitoring means, when the encoding of one block is completed, the code amount of the block and 1
The method is characterized in that a difference of a target code amount per block is calculated, and the difference is added to the stock.

An encoding error A% for the image data
In contrast, A% of the target code amount is the stock.

[0010] A compression code amount control method according to the present invention includes a blocking step of dividing image data into a plurality of blocks of a predetermined pixel size, an orthogonal transformation step of performing discrete cosine transformation of the blocked image data, Storing in a quantization table; a quantization step of dividing the transform coefficient by a coefficient of the quantization table; an encoding step of compressing data output from the quantization means; A code amount monitoring step of monitoring a code amount per block, and a storage step of storing a stock of a spare code amount calculated from a total target code amount of the image data,
The code amount monitoring step compares the target code amount per block with the code amount output from the encoding step, and when the code amount exceeds the target code amount, the code amount of the stock exceeds the target code amount. In the above case, control is performed by adding a part of the stored code amount to use the coding, and the used code amount is subtracted from the stock, and the stored code amount is equal to or less than the excess code amount. In this case, the encoding of the block currently being encoded is forcibly terminated.

Further, the code amount monitoring step calculates a difference between a code amount of the block and a target code amount per block when encoding of one block is completed, and adds the difference to the stock. It is characterized by the following.

[0012]

Embodiments of the present invention will be described with reference to the drawings.

The present invention orthogonally transforms image data of one screen, quantizes the transform coefficients, and sequentially monitors the amount of code generated in the encoding process for each block in the compression of a still image to be encoded. Provided is an image compression encoding apparatus that, when the code amount allocated to one block is exceeded, allocates a code amount secured in advance, or terminates the encoding to record a desired code amount on a recording medium. Is what you do. In a process of compressing image data with an error within a predetermined range, as a pre-process for creating a quantization table of the present process, using block, orthogonal transform, a reference quantization table, and performing quantization and encoding I do.

Next, a quantization table for this processing is created from the code amount compressed using the reference quantization table, and this processing is performed. In the encoding of the present process, the target code amount per block is calculated by subtracting the stock acting as the spare code amount from the total target code amount and dividing by the total block number (Equation 1).
And What is important here is that the stock is used when the stock needs an error within a predetermined range equal to or more than the target code amount per block.

During the encoding process, the code amount is monitored each time one-component encoding is performed. If the target code amount per block is exceeded, the stock is used if the stock is larger than the excess code amount. Is performed, and if the stock is insufficient, EOB (End Of Block) is inserted and the encoding of the block is completed. The block is encoded to the end, and if the code amount is less than the target code amount per block, the difference is added to the stock. Thus, the target code amount per block is calculated, and 1
The image data is compressed to a desired code amount by monitoring the code amount per block. Target code amount per block = (Target code amount−Stock) / Total number of blocks (Equation 1) Referring to FIG. 1, the system configuration of the present invention is as follows.
Blocking means 100 for dividing image data into a plurality of blocks of the same size, for example, 8 × 8 pixels, orthogonal transform means 101 for orthogonally transforming the blocked image data, for example, discrete cosine transform, and orthogonal transform unit A quantizing means 102 for dividing the output transform coefficient by a coefficient shown in the quantization table 104; an encoding means 103 for compressing data output from the quantizing means by a redundancy compression coding method such as Huffman coding And a code amount monitoring means 106 for monitoring the code amount generated by the coding means. Here, as described below, the code amount monitoring means 106 monitors the code amount, and uses the stock for the storage means (not shown) according to the code amount. It controls whether to suspend or add to the stock.

Next, the overall operation of this embodiment will be described in detail with reference to the flowcharts of FIGS. As a pre-process for creating a quantization table for this process, compression using a reference quantization table is performed. FIG.
, The image data is divided into 8 × 8
It is divided into a plurality of blocks each having a pixel size (S10 in FIG. 2).
0), the orthogonalized transformation means 10
The discrete cosine transform is performed by 1 (S101), the transform coefficient is divided by the coefficient shown in the reference quantization table 104 by the quantization means 102 (S102), and the data output from the quantization means is encoded by the encoding means 105. Data is compressed by a redundancy compression coding method such as Huffman coding (S10).
3). The coefficient of the quantization table for this processing is calculated from the code amount of the compressed data (S104), and this processing is started. In this processing, when encoding is performed within a predetermined range with respect to the target code amount, for example, an error within 10%, a value obtained by dividing 90% of the target code amount by the total number of blocks and subtracting EOB for each block is used. (S10)
5) 10% of the target code amount is used as a stock when a code amount more than the code amount per block is required during encoding. This stock is set in stock storage means (not shown).

After performing blocking and orthogonal transformation in the same manner as in the preprocessing, quantization is performed using the quantization table for this processing (S106). In the encoding of this processing (S107),
In this processing, after the DC component (DC component) of the upper left coefficient of the 8 × 8 matrix is encoded (S200 in FIG. 3), the AC component (AC component) of the other coefficient is encoded once (S201). The code amount of the block is monitored using the code amount monitoring means 106 (S202).

If the target code amount per block exceeds the target code amount during encoding, if the code amount stored is equal to or greater than the code amount exceeded, the code amount monitoring means 106 stores the code amount in the stock storage means. Encoding is performed using the code amount, the used code amount is subtracted from the stock (S204), and the next AC component is encoded. If the stored code amount is equal to or less than the excess code amount, EOB is inserted to forcibly end the encoding of the block currently being encoded (S205).

By inserting the EOB, high-frequency terms thereafter are cut.

If all the AC components in one block are coded, the difference between the code amount of the block and the target code amount per block is added to the stock (S208), and the difference between the block and the block that does not require a large code amount is added. The coding amount is allocated to a block requiring a large amount, and coding of the next block is started. Since the stock increases or decreases, it is possible to allocate the code amount of a block that does not require a code amount to the block that requires a large code amount. Also,
After encoding the entire screen, if there is any remaining stock, the remaining stock is replaced with the stock initial value (10% of the target code amount).
, And re-executing the coding, it is possible to allocate a code amount to the block in which the EOB is inserted because the stock is insufficient, and it is possible to encode a high-frequency term, thereby improving the image quality. Further, in the present invention, when calculating the code amount per block, instead of dividing the target code amount by the total number of blocks, a means for storing the code amount of each block in the preprocessing and the compressed code amount is provided. By adding, the code amount compressed for each block is divided by the code amount compressed in the preprocessing,
By multiplying the target code amount (Equation 2), it is possible to calculate a detailed block allocation code amount. Target code amount per block = (Target code amount−Stock) × (Code amount of pre-processing block / Code amount of pre-processing)
(Equation 2)

[0021]

According to the present invention, encoding can be performed most efficiently within a limited code amount. The reason is that a small code amount is allocated to a block having a small code amount, and the remaining code amount is allocated to a block requiring a large code amount, thereby enabling efficient coding and improving image quality. is there.

Further, even if there is some error in the quantization table, it can be compressed to the target code amount. The reason is that, since the code amount per block is monitored, the same effect as changing the coefficient of the quantization table by cutting the high-frequency terms can be obtained.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a compression code amount control device for explaining a first embodiment of the present invention.

FIG. 2 is a flowchart for explaining the present invention.

FIG. 3 is a flowchart for explaining the present invention.

FIG. 4 is a block diagram showing a compression code amount control device for explaining a conventional example.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 100 Blocking means 101 Orthogonal transformation means 102 Quantization means 103 Encoding means 104 Quantization table 106 Encoding monitoring means 200 Blocking means 201 Orthogonal transformation means 202 Quantization means 203 Encoding means 204 Quantization table

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Hiroaki Matsumoto 1-403, Kosugi-cho, Nakahara-ku, Kawasaki-shi, Kanagawa 53 F-term in NEC Ic Microcomputer System Co., Ltd. 5C059 KK22 KK26 MA00 MA23 MC14 ME02 PP01 TA47 TA49 TB08 TC18 TC19 TD12 UA02 UA38 5C078 AA04 BA57 CA34 CA35 DA07 DA11 DB07 9A001 BB02 EE02 EE05 FF05 GZ01 HH27

Claims (6)

[Claims] A block unit for dividing the image data into a plurality of blocks each having a predetermined pixel size; an orthogonal transform unit for performing a discrete cosine transform on the block image data; a quantization table for storing coefficients; Quantizing means for dividing the transform coefficient from the orthogonal transform means by the coefficient of the quantization table; coding means for compressing data output from the quantization means; and a code amount per block of the coding means And a means for storing a stock of a preliminary code amount calculated from the total target code amount of the image data, wherein the code amount monitoring means determines a target code amount per block and the target code amount. The code amount of the output of the encoding means is compared, and when the code amount of the stock exceeds the target code amount, if the code amount of the stock is equal to or greater than the excess code amount, the encoding means is Control is performed by adding a part of the stored code amount in the stock storage means and using coding, and the used code amount is subtracted from the stock, and the stored code amount is equal to or less than the excess code amount. Wherein the encoding of the block currently being encoded is forcibly terminated. 2. The code amount monitoring means further comprising, when encoding of one block is completed, the code amount of the block
2. The compression code amount control device according to claim 1, wherein a difference between the target code amounts per block is calculated, and the difference is added to the stock. 3. An encoding error A for the image data.
2. The compression code amount control device according to claim 1, wherein A% of the target code amount is the stock with respect to%. 4. A step of dividing image data into a plurality of blocks of a predetermined pixel size, an orthogonal transformation step of performing discrete cosine transformation of the blocked image data, and a step of storing coefficients in a quantization table. A quantization step of dividing the transform coefficient by a coefficient of the quantization table, an encoding step of compressing data output from the quantization means, and monitoring a code amount per block by the encoding step. A code amount monitoring step; and a storing step of storing a stock of a preliminary code amount calculated from a total target code amount of the image data. The code amount monitoring step includes the step of: The code amount of the output of the process is compared, and when the code amount of the stock exceeds the target code amount, if the code amount of the stock is equal to or larger than the code amount of the excess, one of the stock code amount is compared. Is added, and the used code amount is subtracted from the stock.If the stocked code amount is equal to or less than the excess code amount, the coding of the block currently being coded is performed. A compression code amount control method, wherein the compression code amount control is forcibly terminated. 5. The code amount monitoring step further includes, when encoding of one block is completed, the code amount of the block being
5. The compression code amount control device according to claim 4, wherein a difference between target code amounts per block is calculated and the difference is added to the stock. 6. An encoding error A for said image data.
5. The compression code amount control method according to claim 4, wherein A% of said target code amount is said stock with respect to%.