JP2000261805A - Picture compression system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a picture compression system where a degree of compression can be adjusted in detail with high precision at high speed. SOLUTION: The picture compression system is provided with a discrete cosine transform means 2 that applies discrete cosine transform to luminance data and color difference data in the unit of blocks to generate DCT coefficients in the unit of blocks, code data generating means 4, 5 that sequentially generate respective code data with 1st and 2nd compression degrees set to the luminance data and with 1st and 2nd compression degrees set to the color difference data on the basis of the DCT coefficients, and compression degree estimate means 6, 7 that estimate a luminance data compression degree or a color difference data compression degree to generate code data with an object code data quantity for the luminance data or the color difference data in response to the respective code data quantity of the 1st 2nd compression degrees for the luminance data or the color difference data.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing system, and more particularly to an image compression system capable of compressing an image to reduce the amount of data.

[0002]

2. Description of the Related Art One of those using an image compression system is a digital still camera. A digital still camera shoots a digital still image by pointing a lens at a subject and pressing a shutter button. The image formed via the lens is converted into an electric signal, compressed, and stored in a replaceable memory card or the like. Data compression is a process for reducing the amount of data in order to store a large amount of image data in a memory card.

[0003] The amount of code data obtained by data compression of a digital image differs depending on the spatial frequency distribution and the like of the digital image. For example, for a digital image containing many high-frequency components, the amount of code data cannot be reduced much. On the other hand, with respect to a digital image having few high frequency components, the amount of code data can be considerably reduced. That is, generally, the amount of code data generated by data compression differs depending on the type of digital image.

[0004] Code data obtained by data compression is stored in a storage medium such as a memory card. The memory card has, for example, a storage capacity of 1 Mbyte. In that case, data of 1 Mbyte or more cannot be stored in the memory card.

[0005] It is desirable to notify the photographer of the remaining recordable number in order to avoid writing code data exceeding 1 Mbyte in the memory card or for the convenience of the photographer. If the code data to be data-compressed has the same data amount for each image regardless of the type of digital image, the number of digital images that can be recorded on the memory card can be easily notified to the photographer.

However, when the amount of code data differs for each image, it is difficult to notify the photographer of the remaining number. If the code data amount of the image to be captured is small, a large number of images can be recorded. If the code data amount of the image to be captured in the future is large, only a small number can be recorded.

When the fixed length processing of the code data is performed, any kind of digital image can be converted into a substantially constant amount of code data. The fixed length processing is performed for one sheet (1
This is a process for compressing data of a digital image of a frame to generate fixed-length code data. If the code data has a fixed length, the number of remaining images can be easily notified to the photographer.

Next, the fixed length processing will be described. To perform the fixed length processing, first, statistical processing is performed as preprocessing, and the degree of data compression is adjusted according to the result of the statistical processing to generate fixed-length code data.

When a photographer presses a shutter button, a digital image is captured by a digital still camera. next,
Statistical processing is performed on the captured digital image.
Statistical processing is processing for estimating how much code data is generated when compression is performed on a captured digital image.

When the statistical processing is completed, compression processing and storage processing are performed. As a result of the statistical processing, if it is estimated that a large amount of code data is likely to be generated, compression may be performed by setting a higher compression degree. If it is estimated that the code data is likely to be generated in a small amount, the compression may be performed by setting the compression degree to a low level. Code data generated by data compression always has a substantially constant data amount.

[0011] After that, the code data that has been subjected to the data compression by the storage process is recorded on the memory card. Above,
A series of processes from capture of the digital image to recording on the memory card ends.

[0012]

In the image compression processing, there is the following method for performing the fixed length processing. First, as statistical processing, compression processing is performed once at a reference compression degree. As a result, when a data amount larger than the target data amount is generated, the compression degree is set to be higher than the reference compression degree. On the other hand, when the amount of code data smaller than the target data amount is generated, the compression degree is set lower than the reference compression degree. Formal compression processing is performed at the set compression degree, and code data of an image is generated.

However, the relationship between the degree of compression and the amount of code data is not constant, and differs depending on the type of image. Therefore, an error occurs between the target data amount and the generated code data amount. The accuracy of the fixed length is low only by performing the statistical processing once.

In order to perform a more precise fixed length processing,
There is a method in which statistical processing is performed twice or more to determine the degree of compression. The case where the statistical processing is performed twice will be described as an example. First,
As the first statistical processing, the compression processing is performed at the first reference compression degree, and the code data amount is estimated. Next, as the second statistical processing, the compression processing is performed at a second reference compression degree different from the first reference compression degree, and the code data amount is estimated. A formal compression degree is set in consideration of the above two code data amounts. For example, when the target data amount is between the two code data amounts, a formal compression degree may be set between the first reference compression degree and the second reference compression degree. Formal compression processing is performed at the set compression degree, and code data of an image is generated.

However, in this method, since the statistical processing is performed twice and the formal compression processing is performed once, the compression processing is performed three times in total, which requires a lot of time.

Furthermore, in the case of a color image, luminance data may be emphasized depending on the required image quality, and conversely, color data (color difference data) may be emphasized.
For example, in a digital still camera having a large number of pixels such as 1.3 million pixels, the influence of luminance data is larger than that of chrominance data as compared with a digital still camera of 350,000 pixels having a smaller number of pixels. Come. Therefore,
Within the limit of the given amount of compressed image data, finer control of the degree of data compression is required according to the required image quality.

An object of the present invention is to provide an image compression system that performs fixed length processing at high speed and with high accuracy and that can finely adjust the degree of compression.

[0018]

According to one aspect of the present invention, luminance data and chrominance data of image data are divided into a plurality of blocks, and the luminance data block and the chrominance data block are divided into a plurality of blocks. Image data supply means for supplying, discrete cosine transform means for performing discrete cosine transform of a block of luminance data and a block of color difference data supplied from the image data supply means to generate DCT coefficients in block units, and the discrete cosine transform When the DCT coefficient of one luminance data block is input from the means, code data is sequentially generated for the DCT coefficient at the first and second compression degrees set for luminance data,
When the discrete cosine transform means generates a DCT coefficient of one color difference data block, the DCT coefficient
Code data generation means for sequentially generating respective code data at the third and fourth compression degrees set for color difference data, and first and second compression for the luminance data generated by the code data generation means The amount of code data for each of the degrees is accumulated for all supplied luminance data blocks, and the codes for the third and fourth compression degrees for the chrominance data generated by the code data generating means are calculated. A counter that accumulates the amount of data for all supplied chrominance data blocks, and a luminance value according to the respective code data amounts of the first and second compression degrees for the luminance data generated by the counter. Estimating the luminance data compression degree for generating code data of the target code data amount of the data, and calculating the color data for the color difference data generated by the counter. And a compression degree estimating means for estimating a color difference data compression degree for generating code data of a target code data amount of the color difference data according to the respective code data amounts of the fourth compression degree. Is provided.

When the discrete cosine transform means has generated the DCT coefficients of the first block, the code data generating means:
For the DCT coefficient, code data is generated at a first compression degree, and subsequently, code data is generated at a second compression degree. Here, for the luminance data, code data is generated at the first and second compression degrees for luminance data, and for the chrominance data, code data is generated at the third and fourth compression degrees for the chrominance data.

The discrete cosine transform means can generate the DCT coefficients of the first block and then generate the DCT coefficients of the second block. When the discrete cosine transform means has finished generating the DCT coefficients of the first block, the code data generating means generates code data with at least two types of compression for the luminance data or the chrominance data; The process of generating the DCT coefficients of the two blocks can be parallelized. By performing the steps in parallel, the speed of the compression processing can be increased. Further, by generating code data with at least two types of compression degrees for luminance data and color difference data, it is possible to perform highly accurate and various fixed length processing.

[0021]

FIG. 2 is a flowchart showing a processing procedure performed by an image compression system according to an embodiment of the present invention.

In step S1, statistical processing is performed. In this statistical processing, the amount of code data of the code data obtained by coding the image data using two kinds of compression degrees during one compression time is calculated. Usually, in order to calculate the code data amount of code data encoded using two types of compression degrees, at least two compression processing times are required. The feature of the statistical processing here is that code data encoded using two kinds of compression degrees is efficiently generated and processed in a short time. After generating the code data, a formal degree of compression is determined based on the amounts of the two types of code data.

In step S2, formal compression processing is performed at the compression degree determined by the statistical processing. By this compression processing,
Code data is generated. The amount of code data generated is close to the target data amount. Thus, the fixed-length compression processing ends.

A color image has luminance data (Y), blue difference data (Cb), and red difference data (Cr). The blue color difference data and the red color difference data are collectively referred to as color difference data.

There are various data formats for a color image. In the [4: 2: 0] format, the ratio of the number of luminance data, the number of blue difference data, and the number of red difference data is 4: 1: 2.
It is one. In the [4: 2: 2] format, the ratio between the number of luminance data, the number of blue difference data, and the number of red difference data is 2: 1: 2.
It is one.

FIG. 1 is a block diagram showing the configuration of the image compression system according to the present embodiment. This image compression system uses JPE, a standard compression method for digital still images.
Generate G (joint photographic expert group) type code data. The resources of the conventional JPEG system can be used as they are.

The image compression system includes an image memory 1, a discrete cosine transform (hereinafter referred to as DCT) unit 2, a DCT coefficient memory 3, a quantization unit 4, an encoding unit 5, a code data amount counting unit 6, a scale factor determining unit. 7 and a controller 8. The controller 8 exchanges timing signals with all other processing blocks and adjusts timing between the processing blocks.

Next, each processing block will be described.

The image memory 1 is, for example, a DRAM or a flash memory, and stores one frame of image data. The image memory 1 stores image data in a normal raster format. The image data is composed of two-dimensional pixel data.

The raster format is a format in which the following pixel data of one frame image are arranged in the following order. First, the pixel data of the pixels starting from the pixel at the upper left corner of the image and continuing in the right horizontal direction are sequentially arranged. After reaching the rightmost pixel, the pixel data of successive pixels starting from the leftmost pixel of the next line and continuing in the right horizontal direction are sequentially arranged. Hereinafter, similarly, the pixel data of the pixels up to the bottom line are arranged. The pixel data of the pixel at the lower right corner is the last data.

Since the image compression system basically performs processing for each block of 8 × 8 pixels, the image memory 1 converts image data from a raster format to a block format,
Supply to T section 2. A monochrome image has one type of image data. A color image is divided into luminance data (Y), blue color difference data (Cb), and red color difference data (Cr), and raster / block conversion is performed using each of them as separate image data.

The block format is an arrangement of the following pixel data for one frame image. One frame image is divided into a plurality of blocks. One block is composed of 8 × 8 pixels. The blocks in one frame are arranged in the same manner as the above-described order of the pixel data in the raster format. The last block is
This is the block in the lower right corner. The arrangement of the pixel data in the block is also the same as the arrangement of the pixel data in the raster format, starting from the pixel data at the upper left corner in the block and lining up in the right horizontal direction. The last pixel data is the pixel data at the lower right corner in the block.

The DCT unit 2 is supplied with image data I in block format. For example, in the [4: 2: 0] format, four blocks of luminance data are supplied first, followed by one block of blue difference data, and then one block.
Red difference data for the block is provided. Hereinafter, the image data is supplied in block units by repeating this order.

Hereinafter, processing is performed using one block of luminance data or one block of color difference data as one unit. That is, in the JPEG compression, one image is divided into 8 × 8 pixel blocks, and the following processing is performed for each block.

The DCT unit 2 performs a DCT process on the image data I in block units. In the DCT processing, the DCT coefficient F is obtained by sandwiching the image data I between a transposed cosine coefficient matrix ^Dt and a cosine coefficient matrix D and performing a matrix operation as in the following equation.

F = D ^t ID Here, the DCT coefficient F is an 8 × 8 matrix and indicates a spatial frequency component.

The DCT coefficient memory 3 is, for example, a DRAM or an SRAM, and has a DCT coefficient F generated by the DCT unit 2.
Is stored.

Next, the configuration of the quantization unit 4 will be described. In the memory 11, an area for the reference quantization table Q is secured.

FIG. 3 shows an example of the reference quantization table Q. As described above, the image compression system performs data compression in units of 8 × 8 blocks, and accordingly, the quantization table Q is also formed of an 8 × 8 matrix.

The reference quantization table Q is a quantization table for performing data compression at a standard compression degree. The quantization process divides the 8 × 8 DCT coefficient F by the corresponding coefficient in the quantization table Q. The elements in the matrix of DCT coefficients have lower spatial frequency components in the upper left direction of the matrix and higher frequency components in the lower right direction. The reference quantization table Q indicates that the lower the frequency component as a whole, the finer the quantization, and the higher the frequency component, the coarser the quantization. Generally, data compression is performed by removing information on high-frequency components of image data in consideration of human visual characteristics and high-frequency components having a lot of noise.

In FIG. 1, a multiplier 12 adds a scale factor SF to all elements of a matrix of a reference quantization table Q.
Multiply by SF × Q is data forming an 8 × 8 matrix, and is hereinafter referred to as a quantization table for convenience of explanation.

The scale factor SF is one of a first scale factor Y_SFa and a second scale factor Y_SFb for luminance data and a first scale factor C_SFa and a second scale factor C_SFb for color difference data. Selected.

The first and second scale factors Y_SFa and Y_SFb for luminance data are the first and second scale factors for color difference data.
And the second scale factors C_SFa and C_SFb.

The multiplier 12 outputs a quantization table SF × Q. The quantization table SF × Q is stored in the memory 13a,
13b or one of the memories 14a and 14b.

The memory 13a stores a first quantization table A1 (Q × Y_SFa) for luminance data, and the memory 13b stores a first quantization table B1 (Q × C_SFa) for color difference data. It is memorized. Also, memory 1
4a includes a second quantization table A2 for luminance data.
(Q × Y_SFb) is stored in the memory 14b, and a second quantization table B2 (Q × C_SF) for color difference data is stored in the memory 14b.
b) is stored. Quantization table A1 for luminance data
And A2 have different scale factors SF. Similarly, the scale factors differ between the color difference data quantization tables B1 and B2.

The degree of compression of the code data is determined by the value of the scale factor SF. That is, the difference between the quantization table A1 for luminance data and the quantization table A2 indicates a difference in the degree of compression. The quantization table A1 for luminance data is Q
× Y_SFa, and the quantization table A2 is Q × Y_
It is represented by SFb. Quantization table B1 for color difference data
Is represented by Q × C_SFa, and the quantization table B2 is Q
× C_SFb.

The divider 15 has quantization tables A1, A
2, either B1 or B2 is supplied as SF × Q. When performing the statistical processing in step S1 of FIG. 2, in the case of luminance data, the quantization table A1 is used as a first reference table, and the quantization table A2 is used as a second reference table. In the case of color difference data, the quantization table B1 is used as a first reference table, and the quantization table B2 is used as a second reference table.

The divider 15 has a DCT as shown in the following equation.
The DCT coefficient Fuv stored in the coefficient memory 3 is divided by a quantization table SF × Quv to output a quantization coefficient Ruv.

[0049]

Ruv = round [Fuv / (SF × Quv)] Here, u and v specify elements in the matrix. Rounding round means rounding to the nearest integer.

The encoding section 5 performs an encoding process on the quantized data Ruv. The encoding process includes processes of run-length encoding and Huffman encoding. Run-length encoding is used for data in which values of 0 are continuously present.
High compression can be performed. In the quantized data Ruv, many zeros are likely to be collected in the lower right part (high-frequency component) of the matrix. By utilizing this property and performing run-length encoding of the matrix Ruv of quantized data by zigzag scanning, high compression can be performed. Zigzag scanning is a method of sequentially scanning from low frequency components to high frequency components.

The encoding section 5 performs Huffman encoding after performing run-length encoding, and generates encoded data.

The code data amount counting section 6 includes a counter A
1 and a counter B1 and a counter A2 and a counter B2. The counter A1 quantizes the luminance data using the quantization table A1, and counts the code data amount NA1 when the code data is generated. The counter A2 quantizes the luminance data using the quantization table A2 and counts the code data amount NA2 when the code data is generated. The counter B1 quantizes the color difference data using the quantization table B1, and counts the code data amount NB1 when the code data is generated. The counter B2 quantizes the color difference data using the quantization table B2 and counts the code data amount NB2 when the code data is generated.

In other words, at the time of the statistical processing, the counter A1 counts the code data amount NA1 when the luminance data is compressed by the first reference compression degree for luminance data (the quantization table A1 for luminance data). . Counter A
2 counts the code data amount NA2 when the luminance data is compressed by the second reference compression degree for luminance data (the quantization table A2 for luminance data). The counter B1 is
The code data amount NB1 when the color difference data is compressed using the first reference compression degree for color difference data (quantization table B1 for color difference data) is counted. The counter B2 counts the code data amount NB2 when the color difference data is compressed using the second reference compression degree for color difference data (the quantization table B2 for color difference data).

The counters B1 and B2 count the total amount of the blue difference data and the red difference data.

FIG. 4 shows the code data amount counted by the code data amount counting section 6. Note that the same applies to the color difference data, and therefore, the case of luminance data will be described as an example. An image of one frame has, for example, luminance data of n blocks. Here, n is any positive integer. Code data is generated in block units. The counters A1 and A2 accumulate the amount of code data of all blocks (n blocks) and calculate the amount of code data of luminance data of one frame image.

The counter A1 calculates a code data amount NA1 when the luminance data is compressed using the quantization table A1. The counter A2 calculates a code data amount NA2 when the luminance data is compressed using the quantization table A2.

Here, an example is shown in which the scale factor Y_SFa for the quantization table A1 is larger than the scale factor Y_SFb for the quantization table A2. In other words, when the quantization table A1 is used, higher compression can be performed than when the quantization table A2 is used.

Code data amount NA1 when the same luminance data is compressed using quantization table A1.
Is smaller than the code data amount NA2 when compression is performed using the quantization table A2.

In FIG. 1, a target code data amount NX and a data ratio RT are input to a conversion unit 9. The data ratio is represented by a luminance code data amount Y_NX and a chrominance code data amount C.
_NX. The conversion unit 9 generates a target luminance code data amount Y_NX and a target color difference code data amount C_NX based on the target code data amount NX and the data ratio RT. The target code data amounts Y_NX and C_NX are supplied to the scale factor determination unit 7.

In FIG. 1, a scale factor determining unit 7
Is based on two types of code data amounts NA1 and NA2,
Determine the scale factor Y_SFx for formal luminance data. The scale factor Y_SFx is a value estimated as a degree of compression for generating code data of a target code data amount Y_NX of luminance data. When the scale factor Y_SFx is obtained, the statistical processing (step S in FIG. 2)
1) ends. Scale factor C_S of color difference data
Fx is determined basically in the same manner as in the case of luminance data. That is, the scale factor C_SFx corresponding to the target code data amount C_NX of the color difference data is obtained based on two types of code data amounts NA1 and NA2 to which different quantization tables B1 and B2 are applied.

The total value of the target luminance data amount Y_NX and the target color difference data amount C_NX is set so as to be a constant value NX or a substantially constant value. The ratio RT between the target luminance data amount Y_NX and the target chrominance data amount C_NX can be changed according to the desired reproduced image quality. That is, the code data amounts Y_NX and C_NX can be set to different values.

After the statistical processing is completed, compression processing (step S2 in FIG. 2) is performed. In the compression processing, the scale factor Y_SFx (or C_SFx) is input to the quantization unit 4 as the scale factor SF. If compression is performed using the scale factor Y_SFx (C_SFx), the luminance data (color difference data) is converted to the target data amount Y_NX (C_SFx).
NX).

FIG. 5 shows the scale factor determining unit 7 of FIG.
6 is a graph for explaining the processing of FIG. The vertical axis is the scale factor, and the horizontal axis is the code data amount. Note that the same applies to the color difference data, and therefore, the case of luminance data will be described as an example. The scale factor determination unit 7 holds the scale factor as a function of the amount of code data. As an example, a case where the scale factor and the code data amount have a linear relationship will be described.

When compression is performed using the scale factor Y_SFa, a code data amount NA1 is obtained. When compression is performed using the scale factor Y_SFb, a code data amount NA2 is obtained. Based on these data amounts, characteristics (approximate by a straight line) as shown in FIG. 5 are set.

The target luminance data amount Y_NX is designated externally. The scale factor Y_SFx corresponding to the target data amount Y_NX is obtained from the above-described characteristics.

The scale factor Y_SFx can be obtained from FIG. 5 as follows.

[0067]

## EQU2 ## (NA2-NA1): (Y_NX-NA1) =
(Y_SFa-Y_SFb): (Y_SFa-Y_SF)
x) (NA2-NA1) x (Y_SFa-Y_SFx) =
(Y_NX−NA1) × (Y_SFa−Y_SFb) (Y_SFa−Y_SFx) = (Y_NX−NA1) ×
(Y_SFa−Y_SFb) / (NA2-NA1) Y_SFx = Y_SFa − ｛(Y_NX−NA1) ×
(Y_SFa-Y_SFb) / (NA2-NA1)｝

Note that, using the above equation, the scale factor Y
In addition to obtaining _SFx, the scale factor Y_SFx (C_SFx) may be obtained by the above approximate expression. Also,
In addition to using a calculation formula, the scale factor Y_SFx (C_SFx) may be determined using a look-up table. Furthermore, the relationship between the scale factor and the amount of code data can be set based on an empirical rule or the like, without being limited to a linear relationship, to another suitable relationship such as an inversely proportional or nonlinear relationship. However, it is desirable that the scale factor Y_SFx (C_SFx) can be determined with a short and simple configuration.

The scale factor Y_SFx (C_
When code data is generated by performing a compression process using SFx), the scale factor obtained above may be further multiplied by a safety factor so that the code data amount does not exceed the target data amount.

Next, in the image compression system of FIG.
Each processing procedure will be described separately for statistical processing and compression processing.

First, the statistical processing will be described. In addition,
Since the same applies to the color difference data, the case of the luminance data will be described as an example.

(1) Statistical Processing Statistical processing is performed in two times for luminance data in one compression processing time.
The types of code data amounts NA1 and NA2 are calculated. This is made possible by taking advantage of the overwhelmingly long processing time in the DCT unit 2 (FIG. 1) in the compression processing.

The processing time in the DCT unit 2 is 5
It accounts for ~ 80%. The time required for DCT processing (DCT unit 2) and the time required for quantization processing (quantization unit 4) of one block (8 × 8 pixels) of image data will be compared.

The DCT unit 2 performs the following calculations. Here, a case where a normal DCT algorithm is used is shown. Multiplication 1024 times Addition 896 times

The quantization section 4 performs the following operations. 64 (= 8 × 8) multiplications

As described above, the number of operations performed by the DCT unit 2 is orders of magnitude larger than that of the quantization unit 4, and
Processing time is overwhelmingly long.

FIG. 6 is a timing chart for performing statistical processing by the image compression system of FIG. The horizontal axis indicates time.

The DCT unit 2 performs processing in the order of processing α1, processing α2, and processing α3. The processing α1 is performed in block 1 (1
This is a DCT process performed on image data (luminance data or color difference data) of the (th block). Process α2 is a DCT process performed on image data of block 2 (second block), and process α3 is performed on image data of block 3 (third block).

At time t0, DCT unit 2 starts DCT processing α1 on the image data of block 1. When detecting the end of the DCT processing α1 at time t10, the controller 8 sends the block 2 to the DCT unit 2.
To start the DCT process α2 for the image data. Subsequently, when the end of the DCT process α2 is detected at time t20, the DCT unit 2 is instructed to start the DCT process α3 for the image data of the block 3. Time t
At 30, the DCT process α3 ends. DCT unit 2
Processes image data continuously from the first block to the last block.

Next, the processing of the quantization unit 4 and the encoding unit 5 will be described. At time t10, the controller 8 performs the DCT processing α on the image data I of the block 1.
When the end of the DCT 1 is detected, the quantization unit 4 is instructed to start the quantization process β1A1 for the DCT coefficient F generated by the DCT process α1. In response to the instruction, the quantization unit 4 uses the quantization table A1 to perform the quantization process β1
Perform A1. The quantization process β1A1 is the DC of block 1
This is a process for quantizing the T coefficient F with the quantization table A1.

At time t11, the controller 8
When the end of the quantization process β1A1 is detected, the quantization process β
The encoding unit 5 is instructed to start the encoding process γ1A1 for the quantized data R generated by 1A1.
The encoding process γ1A1 is an encoding process for data using the quantization table A1 for the block 1.

At time t12, the encoding process γ1A1
Is completed, code data is generated by using the quantization table A1. Thereafter, although not shown in FIG. 6, the counter A1 of the code data amount counting unit 6 (FIG. 1)
Counts the amount of the code data.

The above is the quantization process and the encoding process for the quantization table A1. Next, quantization table A
2 and a quantization process and an encoding process.

At time t12, the controller 8
When the end of the encoding process γ1A1 is detected, the DCT process α
1 for the DCT coefficient F generated by
Instruct the quantization unit 4 to start 1A2. However, at this time, the controller 8 instructs to perform the quantization process β1A2 using the quantization table A2. The quantization process β1A2 is a process of quantizing the DCT coefficient F of the block 1 using the quantization table A2.

At time t13, the controller 8
When the end of the quantization process β1A2 is detected, the quantization process β
The encoding unit 5 is instructed to start the encoding process γ1A2 for the quantized data R generated by 1A2.
The encoding process γ1A2 is an encoding process for data using the quantization table A2 for the block 1.

At time t14, the encoding process γ1A2
Is completed, code data resulting from the use of the quantization table A2 is output. Thereafter, although not shown in FIG. 6, the counter A2 of the code data amount counting unit 6 (FIG. 1)
Counts the amount of the code data.

At time t12, quantization table A1
At the time t14, the generation of the code data by the quantization table A2 ends. Thereafter, at time t20, the DCT process α2 ends. That is, while performing the DCT processing α2, the quantization processing β1A1 and β1A2 and the encoding processing γ
1A1 and γ1A2 are sequentially performed.

In the above, the quantization processing and the encoding processing have been performed on the block 1. Next, quantization processing and encoding processing are performed on the block 2.

At time t20, the controller 8
When the end of the DCT process α2 for the image data I of the block 2 is detected, the DT generated by the DCT process α2 is
The quantization unit 4 is instructed to start the quantization process β2A1 for the CT coefficient F. At that time, the controller 8
It is instructed to perform the quantization process β2A1 using the quantization table A1. The quantization process β2A1 is a process of quantizing the DCT coefficient F of the block 2 using the quantization table A1.

At time t21, the controller 8
When the end of the quantization process β2A1 is detected, the quantization process β
The encoding unit 5 is instructed to start the encoding process γ2A1 for the quantized data R generated by 2A1.
The encoding process γ2A1 is an encoding process for data using the quantization table A1 for the block 2.

At time t22, the encoding process γ2A1
Is completed, code data resulting from the use of the quantization table A1 is output. Thereafter, although not shown in FIG. 6, the counter A1 of the code data amount counting unit 6 (FIG. 1)
Counts the amount of the code data.

Next, quantization processing and encoding processing using the quantization table A2 are performed.

At time t22, the controller 8
When the end of the encoding process γ2A1 is detected, the DCT process α
Quantization processing β for the DCT coefficient F generated by
It instructs the quantization unit 4 to start 2A2. However, at this time, the controller 8 instructs to perform the quantization process β2A2 using the quantization table A2. The quantization process β2A2 is a process of quantizing the DCT coefficient F of the block 2 using the quantization table A2.

At time t23, the controller 8
When the end of the quantization process β2A2 is detected, the quantization process β
The encoding unit 5 is instructed to start the encoding process γ2A2 for the quantized data R generated by 2A2.
The encoding process γ2A2 is an encoding process for data using the quantization table A2 for the block 2.

At time t24, the encoding process γ2A2
Is completed, code data resulting from the use of the quantization table A2 is output. Thereafter, although not shown in FIG. 6, the counter A2 of the code data amount counting unit 6 (FIG. 1)
Counts the amount of the code data.

At time t22, the quantization table A1
Is completed, and at time t24, the generation of code data by the quantization table A2 ends. Then, at time t30, the DCT process α3 ends. While performing the DCT process α3, the quantization processes β2A1 and β2A2 and the encoding processes γ2A1 and γ2
A2 ends.

Hereinafter, similarly, the processing is continued up to the last block. The time of this series of statistical processing is determined by the DCT unit 2
Corresponds to the time for performing the DCT processing on all the blocks, that is, the time for the normal one-time compression processing. In the case of a block of color difference data, the quantization tables are B1 and B1.
The processing is basically the same as that of the above-described luminance data except that it is 2.

In the image compression system, during the statistical processing, the quantization table A1 (B1) is used within the time of one compression processing.
And two types of code data for the quantization table A2 (B2) and the code data amount NA1 (NB1) and NA
2 (NB2) can be calculated.

FIG. 6 shows the timing in each processing unit. Next, the processing timing for each block will be described.

FIG. 7 is a timing chart showing the timing for performing the statistical processing in units of blocks. The horizontal axis indicates time and corresponds to the time in FIG.

First, the processing of block 1 will be described. At time t0, DCT processing α1 for the image data of block 1 starts, and at time t10,
The CT processing α1 ends. When the DCT process α1 ends, a code data generation process 51A using the quantization table A1 is performed, and then a code data generation process 51B using the quantization table A2 is performed.

The process 51A includes a quantization process β1A1 using the quantization table A1, and an encoding process γ1A1 of the quantized data generated by the quantization process β1A1. The process 51B includes a quantization process β1A2 using the quantization table A2, and a coding process γ1A2 of the quantized data generated by the quantization process β1A2.

Next, the processing of block 2 will be described. At time t10, DCT processing α2 for the image data of block 2 starts, and at time t20,
The DCT process α2 ends. When the DCT process α2 ends, a code data generation process 52A using the quantization table A1 is performed, and then a code data generation process 52B using the quantization table A2 is performed. The process 52A includes a quantization process β2A1 and an encoding process γ2A1. Process 52
B includes a quantization process β2A2 and an encoding process γ2A2.

As described above, since the DCT processing time is long, two quantization processing and coding processing (for example, 51A and 51B) can be performed during DCT processing (for example, α2) for one block. it can. Thus, statistical processing can be performed substantially twice during one compression processing for one frame image. One statistical process uses the quantization table A1 (B1), and calculates the code data amount NA1 (NB1). Another statistical process uses the quantization table A2 (B2), and calculates the code data amount NA2 (NB2).

In the statistical processing, the code data amount NA1 (NB
After calculating 1) and NA2 (NB2), a formal scale factor Y_SFx (C_SFx) is obtained. When the statistical processing is completed, the scale factor Y_SFx
A compression process is performed using (C_SFx). Next, the compression process will be described.

(2) Compression Processing FIG. 8 is a timing chart for performing compression processing by the image compression system of FIG. The horizontal axis indicates time. Since image data is basically the same processing for luminance data and chrominance data, luminance data will be described as an example.

First, at time t50, the DCT unit 2 performs the DCT processing α1 on the image data of the block 1.
To start. At time t60, the controller 8
When the end of the DCT process α1 is detected, the DCT unit 2 is instructed to start the DCT process α2 for the image data of the block 2. Subsequently, when the end of the DCT process α2 is detected at time t70, the DCT unit 2 is instructed to start the DCT process α3 for the image data of the block 3. At time t80, the DCT process α3 ends.
The DCT unit 2 processes the image data continuously from the first block to the last block.

Next, the processing of the quantization unit 4 and the encoding unit 5 will be described. At time t60, the controller 8 performs the DCT processing α1 on the image data of block 1
Is detected, the quantization unit 4 is instructed to start the quantization process β1 for the DCT coefficient generated by the DCT process α1. At this time, the controller 8 instructs to perform the quantization process β1 using the quantization table of the scale factor Y_SFx.

At time t61, the controller 8
When the end of the quantization process β1 is detected, the encoding unit 5 is instructed to start the encoding process γ1 on the quantized data generated by the quantization process β1. The encoding unit 5
The generation of code data for block 1 is started. At time t62, the encoding process γ1 ends.

The above is the quantization processing and coding processing for block 1. Next, quantization processing and encoding processing are performed on the block 2.

At time t70, the controller 8
When the end of the DCT process α2 for the image data of the block 2 is detected, the DC generated by the DCT process α2 is detected.
The quantization unit 4 is instructed to start the quantization process β2 for the T coefficient. At this time, the controller 8 instructs to perform the quantization process β2 using the quantization table of the scale factor Y_SFx.

At time t71, the controller 8
When the end of the quantization process β2 is detected, the encoding unit 5 is instructed to start the encoding process γ2 for the quantized data generated by the quantization process β2. The encoding unit 5
The generation of code data for block 2 is started. At time t72, the encoding process γ2 ends.

In the same manner, the processing is continued until the last block, and the image data of one frame is compressed. Thus, the statistical processing and the compression processing are completed.

FIG. 9 shows the code data amount counting section 6 in FIG.
Here is another example. The code data amount counting unit 6 calculates the code data amount N for the quantization table A1 for luminance data.
A1 and the code data amount N for the quantization table A2
A2, the code data amount NB1 for the color difference data quantization table B1, and the code data amount NB2 for the quantization table B2 are counted.

The code data counting section 6 includes a counter 61
And registers A1, B1, A2 and B2. The counter 61 controls the amount of code data for the quantization table A1, the amount of code data for the quantization table A2, and the amount of code data for the quantization table B1 in a time division manner under the control of the controller 8 in FIG. And quantization table B
The amount of code data for 2 is counted.

The code data amount of the luminance data quantization table A1 in block units is added to the register A1. When the addition for all the blocks is completed, the code data amount NA1 is stored in the register A1. The code data amount NA1 is determined by the quantization table A1.
Is the code data amount of the luminance data of the image of one frame when is used.

The amount of code data of the quantization table A2 in block units is added to the register A2. When the addition for all blocks is completed, the code data amount NA2 is stored in the register A2.
The code data amount NA2 is the code data amount of the luminance data of one frame image when the quantization table A2 is used.

On the other hand, the register B1 is added with the code data amount of the block unit of the quantization table B1 of the color difference data (including the blue difference data and the red difference data). When the addition for all blocks is completed, the code data amount NB1 is stored in the register B1. The code data amount NB1 is the code data amount of the color difference data of one frame image when the quantization table B1 is used.

The amount of code data in block units for the quantization table B2 for color difference data is added to the register B2. When the addition for all the blocks is completed, the code data amount NB2 is stored in the register B2. The code data amount NB2 is determined by the quantization table B2.
Is the code data amount of the color difference data of the image of one frame when is used.

FIG. 10 shows another example of the quantization table 4 of FIG. In the entire image compression system, parts other than the quantization table 4 have the same configuration as that of FIG.

The quantization table 4 has a reference quantization table memory 11, a multiplier 12, a multiplexer 71, and a divider 15. The memory 11 stores a reference quantization table Q. The multiplexer 71 controls the first scale factor Y_S for luminance data under the control of the controller 8.
Fa and the second scale factor Y_SFb, and one of the first scale factor C_SFa and the second scale factor C_SFb for color difference data,
2 is supplied as a scale factor SF. Multiplier 12
Multiplies the reference quantization table Q by the scale factor SF and outputs Q × Y_SFa or Q × Y_SFb or Q × C_SFa or Q × C_SFb.

When the multiplexer 71 has the scale factor Y
When _SFa is supplied to the multiplier 12, the divider 15 performs the following operation and outputs quantized data Ruv. Rounding round means rounding to the nearest integer,
The DCT coefficient Fuv is a coefficient stored in the DCT coefficient memory 3.

Ruv = round [Fuv / (Y_SF
a × Quv)] The quantized data Ruv is encoded by the encoding unit 5, and thereafter, the code data amount NA 1 is counted by the counter A 1 of the code data amount counting unit 6. Code data amount NA1
Is a code data amount when the luminance data is compressed using the first scale factor Y_SFa.

The multiplexer 71 has the scale factor Y
When _SFb is supplied to the multiplier 12, the divider 15 performs the following operation and outputs quantized data Ruv.

Ruv = round [Fuv / (Y_SF
b × Quv)] The quantized data Ruv is encoded by the encoding unit 5, and thereafter, the code data amount NA 2 is counted by the counter A 2 of the code data amount counting unit 6. Code data amount NA2
Is the code data amount when compressed using the second scale factor Y_SFb. The method for the color difference data is basically the same as the method for the luminance data.

The quantizer 4 differs from the quantizer of FIG.
It has only a memory 11 for storing the reference quantization table Q, and does not have a memory for storing a value obtained by multiplying the quantization table Q by the scale factor SF. The image compression system of FIG. 10 can reduce the memory capacity as compared with the image compression system of FIG. 1, so that the size and cost of the system can be reduced.

Note that the multiplexer 71 is connected to the multiplier 12
After the scale factor SF is supplied to the multiplier 15, the operation is performed by the multiplier 12, and then Q × Y_SFa (Q × C_SFa) or Q × Y_SFb (Q × C_SFb) is supplied to the divider 15. That is, the timing of the operation of the multiplier 12 is delayed by one cushion. It is necessary to adjust the timing delay.

In contrast, the image compression system of FIG.
Quantization table memory 13a, 13b, 14a or 14
Since the data read from b is supplied directly to the divider 15, the adjustment of the timing is easy, and there is no delay in the operation.

Next, an example in which statistical processing is performed only on sample blocks will be described. In the above-described statistical processing, processing was performed on all blocks of the image of one frame. However, since the statistical processing is only for estimating the amount of code data, it is not always necessary to perform the processing for all blocks. Therefore, the processing time can be shortened by performing the processing only on the sample blocks instead of performing the processing on all the blocks.

FIGS. 11A to 11C show sample examples of sample blocks of luminance data. The figure shows a collection of two-dimensional blocks. The hatched blocks are sample blocks.

FIG. 11A shows a checkerboard-shaped sample block. FIG. 11B shows a sample block in the form of a vertical stripe. FIG. 11C shows a sample block in the form of a horizontal stripe. For example, in the case of FIG. 11A, statistical processing is performed on every other block in the order of block 1, block 3, block 5, and so on.

These sample blocks are half the number of all blocks. If the statistical processing is performed only on these sample blocks, the time required for the statistical processing can be reduced to about half. However, the amount of code data calculated by the statistical processing is half of the amount of code data of one frame image. Considering that the code data amount is half,
It is necessary to determine the scale factor Y_SFx.

Similarly, statistical processing can be performed only on the sample blocks of the color difference data.

According to this embodiment, statistical processing can be performed substantially twice within the time of one compression processing for one frame image. In the statistical processing, the optimum scale factor can be obtained at high speed and with high accuracy. As a result, the image compression system can quickly and accurately generate an amount of code data close to the target data amount.

In addition, since the scale factor can be individually determined for the luminance data and the color difference data, optimal compressed data suitable for the required image quality can be obtained within the determined limit of the amount of compressed data.

In the statistical processing, a case has been described in which two types of quantization tables are used for each of the luminance data and the color difference data. However, three or more types of quantization tables may be used for each. However, it is desirable that the quantization processing and the encoding processing be within the range in which the DCT processing time for one block is completed. Normally, if two types of quantization tables are used, they will fit within the DCT processing time. If the types of quantization tables to be used are increased, high-precision statistical processing, that is, a more optimal scale factor Y_SFx (C_SFx) can be obtained.

Further, as a method of adjusting the degree of compression, a method of changing the scale factor has been described. However, a method of changing the quantization table itself regardless of the scale factor may be used, or another method may be used.

Further, by separately setting the scale factor of quantization for luminance data and color difference data,
For example, in the case of image data of a multi-pixel CCD (1600 × 1200 or the like), increasing the degree of compression of luminance data (decreasing the amount of data) tends to reduce the resolution of high importance in reproduction image quality. The effect on resolution is small even if the value is increased. Therefore, even with the same code amount, the ratio of the degree of compression can be appropriately adjusted based on the luminance and the color difference, so that the image quality can be made more suitable for the user. In a multi-pixel CCD, it is preferable to reduce the degree of compression of luminance data and increase the degree of compression of color difference data.

[0139] Some color liquid crystal displays have poor color reproducibility. Since such a color liquid crystal display cannot sufficiently express colors, it is preferable to increase the degree of compression of color difference data and decrease the degree of compression of luminance data.

Although the present invention has been described in connection with the preferred embodiments,
The present invention is not limited to these. For example, it will be apparent to those skilled in the art that various modifications, improvements, combinations, and the like can be made.

[0141]

As described above, according to the present invention,
After the discrete cosine transform means has finished generating the DCT coefficients of the first block, the code data generating means generates code data with two kinds of compression for the luminance data or the chrominance data respectively; The steps of generating the DCT coefficients for the next block can be paralleled. By parallelizing the discrete cosine transform and the code data generation process, the speed of the compression process can be increased. By generating code data with individual compression degrees for luminance data and color difference data, it is possible to perform high-precision and optimal data compression under a predetermined code data amount limit.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of an image compression system according to an embodiment of the present invention.

FIG. 2 is a flowchart illustrating a processing procedure performed by the image compression system according to the embodiment.

FIG. 3 is a diagram illustrating an example of a reference quantization table of FIG. 1;

FIG. 4 is a diagram illustrating a method in which the code data amount counting unit of FIG. 1 counts the code data amount.

FIG. 5 is a graph for explaining processing of a scale factor determination unit in FIG. 1;

FIG. 6 is a timing chart showing the timing of each processing unit where the image compression system performs statistical processing.

FIG. 7 is a timing chart showing a block-by-block timing at which the image compression system performs statistical processing.

FIG. 8 is a timing chart showing a timing at which the image compression system performs a compression process.

FIG. 9 is a diagram illustrating another example of the code data amount counting unit in FIG. 1;

FIG. 10 is a diagram illustrating another example of the quantization unit in FIG. 1;

FIG. 11 shows a sample example of a sample block. FIG.
1 (A) is a checkered sample block, FIG.
(B) is a vertical striped sample block, FIG.
(C) is a diagram showing a sample block in the form of a horizontal stripe.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Image memory 2 Discrete cosine transform (DCT) part 3 DCT coefficient memory 4 Quantization part 5 Encoding part 6 Code data amount counting part 7 Scale factor determination part 8 Controller 9 Transformation part 11 Reference quantization table memory 12 Multiplier 13 Quantum Table A Memory 14 Quantization Table B Memory 15 Divider 61 Counter 71 Multiplexer

 ──────────────────────────────────────────────────続 き Continuing on the front page F term (reference) 5C057 AA01 AA07 CE10 EA02 EA07 EL01 EM09 EM13 EM16 GF05 GJ01 GJ03 GM01 GM08 5C059 KK13 KK22 KK27 MA00 MA23 MC11 MC38 ME02 ME05 PP01 PP16 SS15 UA02 DA07 DA07 DA07 DA07 DA07 DA07 9A001 EE02 EE05 HH27 KK42

Claims (7)

[Claims] An image data supply unit that divides each of the luminance data and the chrominance data in the image data into a plurality of blocks, and supplies the luminance data blocks and the chrominance data blocks; Discrete cosine transform means for performing discrete cosine transform on a block of luminance data and a block of chrominance data supplied from, and generating DCT coefficients in block units; Then, with respect to the DCT coefficients, respective code data are sequentially generated with the first and second compression degrees set for luminance data, and the discrete cosine transform means converts the DCT coefficients of one color difference data block. When the DCT coefficient is generated, each of the DCT coefficients is set at the third and fourth compression degrees set for the color difference data. Code data generating means for sequentially generating code data, and the amount of code data for each of the first and second degrees of compression for the luminance data generated by the code data generating means, A counter that accumulates for each block and accumulates the amount of code data for each of the third and fourth compression degrees for the color difference data generated by the code data generation unit for all supplied color difference data blocks A first for the luminance data generated by the counter
And the respective code data amounts of the second compression degree,
Estimating a luminance data compression degree for generating code data of a target code data amount of the luminance data, and respectively generating third and fourth compression data amounts for the color difference data generated by the counter. And a compression degree estimating means for estimating a color difference data compression degree for generating code data of a target code data amount of the color difference data in accordance with the image compression system. 2. The method according to claim 1, further comprising: generating code data for luminance data based on the luminance data compression degree estimated by said compression degree estimating means; 2. The image compression system according to claim 1, further comprising means for instructing the code data generation means to generate the image data. 3. The code data generating means includes a quantizing means for quantizing a DCT coefficient, wherein the quantizing means performs quantization by using first and second quantization tables for luminance data. To generate code data of the first and second compression degrees for the luminance data, and perform quantization using the third and fourth quantization tables for the color difference data, thereby obtaining the color data for the color difference data. 3. The image compression system according to claim 1, wherein code data of the third and fourth compression degrees are generated. 4. The quantization means multiplies the data in the reference quantization table by first and second scale factors for luminance data, thereby obtaining first and second luminance data.
In the third and fourth quantization tables for color difference data by multiplying the data in the reference quantization table by the third and fourth scale factors for color difference data. 4. The image compression system according to claim 3, wherein the system generates data. 5. The image compression system according to claim 3, wherein said code data generation means further includes coding means for performing Huffman coding after quantizing DCT coefficients by said quantization means. 6. The discrete cosine transform means is means for continuously performing discrete cosine transform on luminance data or chrominance data in block units, and the code data generating means is provided for a block having the discrete cosine transform means. Before completing the discrete cosine transform of the luminance data or the chrominance data, completing the generation of the code data of the first and second compression degrees of the luminance data or the chrominance data for the block immediately before the block. Item 6. An image compression system according to any one of Items 1 to 5. 7. The image compression system according to claim 1, wherein said image data supply unit supplies image data for a part of a plurality of blocks constituting an image of one frame. .