JPH06225153A - Picture coder - Google Patents

Abstract

PURPOSE:To provide the picture coder by which a picture is coded with a code quantity as less as possible while keeping sufficient decoding picture quality. CONSTITUTION:The picture coder is made up of a block processing means 1 sampling a picture and dividing it into an input block comprising plural mXn picture elements (m, n are positive integers), a coding means 2 coding the input block, and a code quantity control means 4 controlling the code quantity in the case of coding by the coding means 2. The picture coder is provided with a picture quality estimate means 3 estimating picture quality when the input block is coded and decoded and the coding means 2 codes the picture so as to allow the picture to change from lower decoding picture quality to higher picture quality gradually while referring to the picture quality estimated by the picture quality estimate means 3 for each prescribed picture area. Then an identification code is inserted to code data when the picture quality reaches prescribed decoded picture quality and the data are coded to have a prescribed code quantity.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a device for coding image information.

[0002]

2. Description of the Related Art For example, in an image editing apparatus, image data to be edited or edited image data is temporarily stored in a primary storage device such as a semiconductor memory. Also, the temporarily stored image data is stored in a secondary storage device such as a magnetic disk device. Conversely, the image data stored in the secondary storage device is reproduced and temporarily stored in the primary storage device. The storage capacity required for the primary storage device and the secondary storage device increases as the image size increases, the resolution increases, and the number of gradations per pixel increases. For example, when a JIS A3 size 16770,000 colors / pixel full-color image is stored at a resolution of 16 pixels / mm, the storage capacity is 96 Mbytes. When the storage capacity is increased in this manner, the price of the storage device is increased, and the storage of image data, and
Alternatively, there arises a problem that the reproduction time becomes long.

Among such problems, as a solution to the problem relating to the primary storage device, it has been considered to highly efficiently encode image data in an editable form to reduce the storage capacity. The following three points are required for such encoding.

To obtain a constant compression rate for each predetermined image unit.

Local encoding / decoding is possible for each predetermined image unit.

A uniform encoding / decoding process is performed.

That is, first, since the storage capacity of the primary storage device is finite, it is necessary to be able to encode every predetermined image unit at a preset compression rate irrespective of the content of the image data ( Request point). In this case, the predetermined image unit needs to be at least a page unit. Further, for image editing as described below, the predetermined image unit needs to be a unit obtained by dividing a page into several image regions. For example, it is a block unit of 8 pixels × 8 lines.

Secondly, in order to enable image editing, it is necessary that each predetermined image unit can be independently encoded / decoded (request point).

Thirdly, since encoding / decoding is performed on a primary storage device having a relatively high storage / reproduction speed such as a semiconductor memory,
It is desirable to be able to process at high speed and at a constant speed (request point).

However, in the conventional image coding apparatus for storage / transmission, it is necessary to suppress the visual redundancy and the statistical redundancy of the image data as much as possible. There was a problem that the compression rate fluctuates due to fluctuations. Further, there is a tendency to introduce advanced coding processing such as referring to peripheral image areas in order to code a certain image area, and it is difficult to independently perform encoding / decoding for each predetermined image division unit. Is. Furthermore, it has been difficult to satisfy the requirements (1) to (3) because the introduction of the adaptive processing causes the amount of calculation required for the encoding / decoding processing to fluctuate significantly according to the fluctuation of the redundancy of each image data.

On the other hand, as a conventional technique satisfying the requirements (1) to (3), there is an image coding apparatus filed by the present applicant as Japanese Patent Application No. 4-328265. The image encoding device described in the specification of the application, as shown in FIG. 14, is a block forming unit that samples an image and divides the image into blocks of m × n pixels (m and n are positive integers). 201, analysis means 202 for analyzing the resolution and gradation feature amount of the pixels in the block, and the ranking of each hierarchy when the pixels in the block are divided into the resolution hierarchy and the gradation hierarchy as analysis results. Based on the order of the determined resolution hierarchy and gradation hierarchy, the important order determination means 203 divides the pixels in the block into the resolution hierarchy and the gradation hierarchy, and outputs the divided layers. 204 and importance ranking determination means 203
And the output of the gradation / resolution information layering means 204 are information source coded, code data is output, and the coding processing is terminated when the preset code amount is reached, and an information source coding means 205 And a code amount counting means 206 for adding the amount of data and outputting to the information source coding means 205 that the preset code amount has been reached.

The operation of the image coding apparatus shown in FIG. 14 will be described. The input image data 207 is sampled and divided into m × n pixel input blocks 208 including a plurality of pixels by the blocking unit 201. Next, the analysis unit 202 analyzes the resolution of the block and the feature amount of the gradation. Gradation / resolution feature amount 209 from analysis means 202
Is supplied to the important order determination means 203. When the pixels in the block are divided into a hierarchy of resolution and a hierarchy of gradations by the important order determination means 203, the ranking of each hierarchy is determined by the analysis result from the analysis means 202 (gradation / resolution feature amount 20).
9). In this case, the hierarchy of the predetermined resolution and the hierarchy of the gradation are set as the first order, and the candidates of the hierarchy of the resolution and the candidates of the gradation of which the input block can be decoded with a preset coding error or less are set to the resolution. Based on the feature amount and gradation feature amount, starting from the resolution hierarchy and gradation hierarchy of the first rank, the ranking is performed toward the resolution hierarchy candidate and the gradation hierarchy candidate. Be seen. In the gradation / resolution information layering means 204, the pixels in the block are divided into a hierarchy of resolution and a hierarchy of gradation according to the rank 210 from the important rank determining means 203 and output. Output 210 of rank 210 and gradation / resolution information layering means 204
11 is encoded by the information source encoding means 205 and output as encoded data 212. Further, the amount of code data 212 is added in the code amount counting means 206.
At this time, in the information source coding unit 205, the coding process is terminated when the target code amount reaches a preset target code amount based on the target code amount reaching signal 213 from the code amount counting unit 206, and is equal to or less than the target code amount. The code data 212 having the data amount of is output.

In the encoding system shown in FIG. 14,
For each block, the characteristics of resolution and gradation are analyzed, and the pixels in the block are divided into a hierarchy of resolution and a hierarchy of gradation independently of that, and the order of important hierarchy is reduced in the sense of reducing coding error. This is a method that is determined from the analysis result, data-source-codes the data of each layer based on the rank, and terminates the coding when the total code amount reaches a preset value. According to this method, the coding error due to the coding truncation is small, and at the same time, the coding is always performed with a constant code amount for each block.

Therefore, the coding redundancy or the image quality of the decoding block fluctuates depending on the magnitude of the visual redundancy and / or the statistical redundancy of the input block.
That is, the code amount is set in advance so that the image quality is sufficient even in the block with the lowest decoded image quality, and the excess code amount is assigned to the other blocks in terms of image quality reproduction. Therefore, there is a problem that the overall coding efficiency does not necessarily become high.

As another conventional technique, for example, 198
The image coding apparatus described in D-159, 9th Annual College of Science, Spring is known. The configuration of the image coding device described in the same document will be described below with reference to FIG. However, for simplicity of explanation, only the luminance component of the luminance / color difference components and only the AC component of the conversion coefficients will be described.

The image coding apparatus shown in FIG. 15 comprises a block forming means 301 for sampling an image and dividing it into blocks of 8 × 8 pixels composed of a plurality of pixels, and means for discrete cosine transforming the pixel values in the block (FIG. Medium DCT means) 30
2 and means 30 for calculating the activity of each block
5, means 306 for estimating a quantization step for a predetermined code amount based on the average activity of n blocks for every n blocks, means 303 for quantizing a transform coefficient at the estimated quantization step, n Means 30 for determining the allocated code amount of each block by the ratio of the activity of each block to the total activity of the block
7 and means 30 for variable-length coding the quantized transform coefficient
4 and means 308 for counting the code amount.

Next, the operation of the image coding apparatus shown in FIG. 15 will be described. The input image data 310 is sampled and divided by a blocking unit 301 into an input block 311 of 8 × 8 pixels including a plurality of pixels. The pixel values in the divided input block 311 are discrete cosine transform means 30.
The discrete cosine transform is performed by 2. The above conversion is performed on n blocks. Independently of this, after the activity of each block is calculated by the activity calculating means 305, n is calculated by the quantization step estimating means 306.
Based on the average activity of the blocks, a quantization step having a preset code amount is estimated every n blocks. Further, the code amount allocation unit 307 allocates the code amount 317 of each block according to the ratio of the activity of each block to the total activity of n blocks.
Is determined. On the other hand, the transform coefficient 312 obtained by the discrete cosine transform unit 302 is quantized by the quantization step estimated by the quantization unit 303 to be a quantization index 313, and then the variable length coding unit 304 performs variable length coding. Be converted. At this time, the code amount counting means 308 obtains the total code amount for each block, and when the code amount reaches the assigned code amount, the assigned code amount reaching signal 318 is reached.
Is output to the variable length coding means 304, and the coding of the block is completed.

This system is based on the JPEG baseline system, and the relation between the code amount and the quantization step of the transform coefficient, which is a parameter that influences the code amount, is investigated in advance, and this is used for every n blocks. This is a method of controlling the code amount.

Therefore, in this system, since every n blocks are coded with a constant code amount, as in the prior art described above, it is sufficient even in the image area (n blocks) where the decoded image quality is the lowest. The code amount that provides the image quality is set in advance, and an excessive code amount for image quality reproduction is assigned to the other image area. Therefore, the coding efficiency cannot be made sufficiently high.

On the other hand, in order to solve the problem relating to the secondary storage device, it is said that sufficient decoding image quality can be obtained with a code amount as small as possible under the precondition that image data is temporarily stored in the primary storage device. Coding methods that satisfy the required points are effective.

As shown in FIG. 16, when the conventional technique applied to the primary storage device described above is directly applied to the secondary storage device, that is, the code data temporarily stored in the primary storage device is secondary to the secondary storage device. When storing in a storage device,
Since there is an image area to which an excessive code amount is assigned for image quality reproduction, the required point cannot be satisfied.

[0022]

The prior art shown in FIG. 16 satisfies the above-mentioned requirements (1) to (5) as an encoding device applied to a primary storage device. Since there is an image area to which an excessive code amount is assigned, there is a problem that sufficient decoding image quality and a code amount that is as small as possible cannot be achieved, and the above requirement cannot be satisfied.

Therefore, an object of the present invention is to provide an image coding apparatus capable of coding an image with a code amount as small as possible while obtaining a sufficient decoded image quality.

[0024]

According to the present invention, in order to achieve the above object, an image is sampled and m × n composed of a plurality of pixels.
Blocking means for dividing an input block of pixels (m and n are positive integers), encoding means for encoding the input block, and code amount control for controlling the code amount at the time of encoding in the encoding means. Image quality estimation means for estimating the image quality when the input block is encoded and decoded, and the encoding means estimates the image quality for each predetermined image area by the image quality estimation means. Encoding is performed so that the decoded image quality gradually increases from the lower decoded image quality while referring to the image quality, and when the predetermined decoded image quality is reached, the identification code is inserted into the code data and encoded so that the predetermined code amount is obtained. It is characterized by doing.

The present invention also provides the following object.
An image is sampled and m × n pixels (m, n
Is an integer) input block, an encoding unit that encodes the input block, and a code amount control unit that controls the code amount at the time of encoding in the encoding unit. The encoding device is provided with an image quality estimating means for estimating an image quality when the input block is encoded and decoded, and the encoding means has a predetermined code amount or less and a predetermined decoding amount for each predetermined image area. It is characterized in that encoding is performed so that the image quality is lower than or equal to the image quality, and an identification code indicating that the encoding is completed is added to the code data.

[0026]

The function of the present invention will be described below with reference to specific examples.

FIG. 1 is a block diagram showing the basic configuration of the image coding apparatus of the present invention. The image coding apparatus of the present invention samples the image data 8 and is composed of a plurality of pixels m ×
Blocking means 1 that divides the input block 9 into n pixels (m and n are positive integers), and decoding image quality estimating means 3 that estimates the image quality of the block obtained by encoding / decoding the input block 9 and outputs an estimation result 11 And a predetermined code amount or less for each one or a plurality of blocks, or a predetermined code amount or less,
Moreover, the code amount control means 4 for outputting the encoding parameter 12 so that the image quality is not more than the predetermined decoded image quality, and the input block 9.
Is encoded according to the encoding parameter 12 with a fixed length (maximum code amount limitation), and when the predetermined decoding image quality is reached, the variable length encoding means 2 for inserting the identification code into the code data 10 and the fixed length (maximum code amount limitation) ) Variable by reducing the code data of the image area to which an excessive code amount is allocated for image quality reproduction from the primary storage means 5 for temporarily storing the coded code data 10 and the temporarily stored code data 13. It has a variable length conversion means 6 for long-length encoding and a secondary storage device 7 for storing the variable-length encoded code data 14.

The image data 8 is sampled and divided into m × n pixel input blocks 9 composed of a plurality of pixels by the blocking means 1. Next, by the decoded image quality estimation means 3,
The image quality of the encoded / decoded input block 9 is estimated,
The image quality estimation result 11 is output. The code amount control means 4 refers to the input block 9, the code data 10, and the image quality estimation result 11 so that one or a plurality of blocks have a predetermined code amount or less, or a predetermined code amount or less, Moreover, the encoding parameter 12 is output so that the image quality is not more than the predetermined decoded image quality. Code amount control means 4
In accordance with the encoding parameter 12 output by the input block 9, the input block 9 is set to have a predetermined code amount or less for each one or a plurality of blocks, or a predetermined code amount or less and a predetermined decoding image quality or less. Variable-length coding means 2 performs fixed-length (maximum code amount limitation) coding. At this time, the variable-length coding means 2 inserts the identification code into the code data 10 when the predetermined decoded image quality is reached. After that, the coding process is completed when the coding is continued up to a predetermined code amount, or when the code amount necessary and sufficient for image quality reproduction is reached. The primary storage means 5 temporarily stores the code data 10 that has been encoded with a fixed length (maximum code amount limitation). The fixed-length (maximum code amount limit) code data 13 temporarily stored is variable-length coded by the variable-length conversion means 6 by selecting only the code data up to the position of the identification code. The secondary storage means 7 stores the code data 14 which has been subjected to variable length coding.

[0029]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The features of the present invention will be specifically described below based on embodiments with reference to the drawings.

An embodiment of the present invention, as shown in FIG.
The blocker 1 that samples the image data 8 and divides it into blocks 8 of m × n pixels (m and n are positive integers) composed of a plurality of pixels, and analyzes the resolution and gradation feature amount of the pixels in the block. A gradation / resolution analyzer 21, a gradation / resolution candidate estimator 22 that estimates a gradation / resolution hierarchy that provides a predetermined decoded image quality from the analysis result 27, and a pixel in a block of resolution hierarchy and gradation. When divided into hierarchies, the ranking of each hierarchy is performed based on the gradation / resolution candidate 11 and the important rank determiner 23, and the rank 2 of the hierarchy of the determined resolution and the hierarchy of the gradation
The gradation / resolution information layering device 24 for dividing the pixels in the block into the resolution layer and the gradation layer in accordance with No. 8 and outputting them.
And a code amount counter 25 that outputs a target code amount reaching signal 29 indicating that a predetermined code amount has been reached by adding the code data 10, an important order 28 and hierarchical gradation / resolution information 30. Information source coding is performed to output code data 10, an identification code is inserted into the code data when the layer indicated by the gradation / resolution candidate 11 is reached, and coded when the target code amount reaching signal 29 is input. End the process, or
After inserting the identification code into the code data, the information source encoder 26 that immediately terminates the encoding, the primary storage circuit 5a that temporarily stores the code data 10, and the identification code is detected from the temporarily stored code data 10. Then, an identification code detector 18 that outputs an address control signal 19 for addressing only the code data from the beginning to the identification code to the primary storage circuit 5a, and a secondary storage circuit that stores the addressed code data 13 7a.

In the block diagram shown in FIG. 2, the decoded image quality estimating means 3 is composed of a gradation / resolution analyzer 21 and a gradation / resolution candidate estimator 22, and an importance ranking determiner 23 and a code amount counter 25. To the code amount control means 4, and the gradation / resolution information layering device 24 and the information source coding device 26 to the variable length coding device 2. The identification code detector 18 corresponds to the variable length conversion means 6.

The operation of the embodiment shown in FIG. 2 will be described below.

The image data 8 is sampled and divided by a blocker 1 into an m × n pixel input block 9 composed of a plurality of pixels. The gradation / resolution analyzer 21 analyzes the resolution and gradation feature amount 27 of the input block 9.
The gradation / resolution candidate estimator 22 is a resolution hierarchy candidate and a gradation hierarchy candidate capable of decoding the input block 9 with a predetermined coding error or less. The gradation / resolution candidate 11 having a predetermined decoded image quality is estimated. Importance determination device 23
In, in the case where the pixels in the block are divided into the resolution hierarchy and the gradation hierarchy, the ranking of each hierarchy is performed based on the gradation / resolution candidate 11. In this case, the hierarchy of the predetermined resolution and the hierarchy of the gradation are set as the first order, and the hierarchy of the resolution and the hierarchy of the gradation are started from the hierarchy of the resolution and the gradation of the first order. The ranking is performed toward. The details of ranking will be described later. In the gradation / resolution information layering device 24, the pixels in the block are divided into a resolution layer and a gradation layer in accordance with the order 28, and are output. In the code amount counter 25, the code data 10 is added, and when the predetermined code amount is reached, the target code amount reaching signal 29 is output. In the information source encoder 26, the important order 28 and the layered gradation / resolution information 30 are encoded, the code data 10 is output, and at the time when the layer indicated by the gradation / resolution candidate 11 is reached. The identification code is inserted into the code data and the encoding process is terminated when the predetermined code amount is reached, or the encoding is terminated immediately after the identification code is inserted into the code data. The code data 10 is temporarily stored in the primary storage circuit 5a. The identification detector 18 detects the identification code from the temporarily stored code data 10, and only the code data from the beginning to the identification code is addressed to the primary storage circuit 5a by the address control signal 19. The addressed code data 13 is stored in the secondary storage circuit 7a. With the above, all the encoding processing for one input block 9 is completed, and the processing for the next input block 9 is started. This is performed for all data of the image data 8.

FIG. 3 is a detailed block diagram of the gradation / resolution analyzer 21. The tone / resolution analyzer 21 includes a tone analyzer 31 that analyzes the tone feature amount of the input block 9 and a resolution analyzer 3 that analyzes the resolution feature amount of the input block 9.
2, and a multiplexer 33 that multiplexes the gradation analysis result 34 output by the gradation analyzer 31 and the resolution analysis result 35 output by the resolution analyzer 32 and outputs the analysis result 27.

The gradation analyzer 31 obtains the distribution of pixel values in the input block 9, for example, and calculates this by the gradation analysis result 3
Set to 4. Alternatively, the difference between the maximum pixel value and the minimum pixel value of the pixel values in the input block 9 may be obtained, and combined with the variance, these may be used as the gradation analysis result 34.

The resolution analyzer 32 uses an image signal analysis method as described below. The details of this image signal analysis method are described in the specification of Japanese Patent Application No. 3-202129 filed by the present applicant, but will be briefly described below.

As shown in FIG. 4, the resolution analyzer 32 is
A block divider 41 that divides the input block 9 by a positive integer ratio j, and an average value separation that calculates the average value of the pixels in the divided block 52 and subtracts this average value from each pixel in the block 52 and outputs 42 and average value separation block 53
And a representative shape block 54 stored in the first vector set 44, a first inner product calculator 43 that calculates the inner product of positive and negative signs of the inner product, and a first vector that stores the previously determined representative shape block Of the set 44, the first index holder 45 that holds the positive and negative signs 55 of the inner product as an index, the average value separation block 53, and the representative shape block stored in the second vector set 47,
Representative shape block 57 indicated by index 56
A second inner product calculator 46 that calculates the inner product of and the positive and negative signs 58 of the inner product, a second vector set 47 that stores a set of representative shape blocks that are obtained in advance, a positive and negative sign 58 of the inner product, and The second index holder 48 that holds the index 56 as a new index, and the representative shape block 60 indicated by the index 59 among the representative shape blocks stored in the average value separation block 53 and the third vector set 50 A third inner product calculator 49 that calculates the inner product and outputs the positive and negative sign 61 of the inner product;
A third vector set 50 for storing a set of representative shape blocks obtained in advance, a positive / negative sign 61 of the inner product and an index 59 are held as new indexes, and the third index holder is output as the resolution analysis result 35. It consists of 51.

The operation of the resolution analyzer 32 will be described below with reference to FIG.

When the resolution feature value is analyzed by the resolution analyzer 32, first, a plurality of representative shape blocks are arranged in each branch of a plurality of stages (three stages in this embodiment) of a binary tree. At this time, the representative shape block is a block having a typical shape, such as a block in which the density gradient direction of the block is the vertical direction, the horizontal direction, or the diagonal direction. Next, the representative shape block arranged in each branch is treated as vector data, the vector difference between two representative shape blocks arranged in a pair of branches is calculated, and the result is used as a difference representative vector to pair the binary tree. Place it in the node where the branch becomes. Finally, the sets of differential representative vectors arranged in the nodes of each stage are collectively stored in the respective vector sets 44, 47, 50.

The input block 9 is divided by the block divider 41 at a positive integer ratio j. The average value separator 42 calculates the average value of the pixels in the divided block 52, and outputs the average value separation block 53 in which the average value is subtracted from each pixel in the divided block 52. First, the first inner product calculator 43 causes the average value separation block 5
The inner product of 3 and the representative shape block 54 stored in the first vector set 44 is calculated, and the positive / negative sign 55 of the inner product is output. The positive / negative sign 55 of the inner product is held by the first index holder 45 as an index. Next, the second inner product calculator 46 causes the average value separation block 5
3 and the representative shape block 57 indicated by the index 56 among the representative shape blocks stored in the second vector set 47, the inner product is calculated, and the positive and negative sign 58 of the inner product is output. The positive and negative signs 58 and the index 56 of the inner product are held by the second index holder 48 as new indexes. Finally, by the third inner product calculator 49, among the representative shape blocks stored in the average value separation block 53 and the third vector set 50,
Representative shape block 60 indicated by index 59
The inner product of and is calculated, and the positive / negative sign 61 of the inner product is output. The sign 61 and the index 59 of the inner product are
The third index holder 51 as a new index
Held by.

In this way, the same processing is performed on each of the divided blocks 52 divided by the division ratio j of the input block 9, and finally the index set for each divided block 52 is held as the third index. Bowl 51
The index set held by is output as the resolution analysis result 35. At this time, the index 35 held in the third index holder 51 represents the shape of the waveform when the input block 9 is regarded as a two-dimensional waveform of the image.

FIG. 5 is a detailed block diagram of the gradation / resolution candidate estimator 22. The gradation / resolution candidate estimator 22 divides the analysis result 27 from the gradation / resolution analyzer 21 into a gradation analysis result 66 and a resolution analysis result 68 and outputs them.
2 and a gradation candidate estimator 63 that obtains a gradation hierarchy required for decoding with a preset encoding error or less based on the gradation analysis result 66 and outputs this as a gradation candidate 67.
A resolution candidate estimator 64 that obtains a hierarchy of resolutions required for decoding with a preset encoding error or less based on the resolution analysis result 68 and outputs this as a resolution candidate 69.
The gradation candidate 67 and the resolution 69 are multiplexed and output as the gradation / resolution candidate 11.

When the gradation candidate estimator 63 obtains the gradation candidate 67 based on the gradation analysis result 66, for example, when the gradation analysis result 66 is the variance in the input block 9, the larger the variance, the more the gradation becomes. The number of layers is increased, and the smaller the number, the smaller the number of layers of gradation. When the gradation analysis result 66 includes the dynamic range (difference between the maximum pixel value and the minimum pixel value) in the input block 9, the larger the variance is, the more the gradation layers are. If the dynamic range is large even if the variance is small, the number of gradation layers is increased.

When the resolution candidate estimator 64 obtains the resolution candidate 69 based on the resolution analysis result 68, it corresponds to a sampling cycle capable of reproducing the waveform according to the resolution analysis result 68, that is, the shape of the waveform of the input block 9. A layer of resolution is assigned and this is set as a resolution candidate 69. For example, when the cycle of the tone change of the waveform shape of the input block 9 is long, that is, when a high spatial frequency component is not included, the sampling cycle is set long, that is, a layer of low resolution is used to reproduce the waveform. Allocate. Further, when the input block 9 has a stepwise gradation change (edge), that is, even a high spatial frequency component is included, in order to reproduce the waveform, the sampling period is shortened, that is, a layer of high resolution is set. Allocate.

In FIG. 2, the gradation from the decoded image quality estimating means 3 /
The resolution candidate 11 is referred to by the rank determining unit 23, and the rank determining unit 23 outputs rank information 28. At this time, starting from a predetermined number of gradations and resolution, the gradation candidates 67
Then, a ranking is determined such that the number of gradations and the resolution are sequentially increased toward the resolution candidate 69, and this is used as ranking information 28.

An example of order determination will be described with reference to FIGS. 6 (a) and 6 (b). The horizontal axis of the graph is the pixel thinning rate, which corresponds to the resolution. The vertical axis is the number of bits per pixel,
This is a value obtained by taking Log ₂ of the number of gradations. The points described as candidates in the graph represent the gradation candidates 67 and the resolution candidates 69. First, a point described as a preset first approximation is used as a starting point. Next, from the starting point to the candidate point,
Through the approximation, the third approximation to the fifth approximation, the rank is sequentially determined up to the candidate point. Here, when the candidate point is reached,
Information indicating that is added to the ranking information 28. After reaching the candidate point, the order is similarly determined sequentially toward the point marked as the original image in the graph. In the example of FIG. 6, the first approximation point is the point with the least image information. FIG. 6A shows the order determining step when the slope of the straight line connecting the first approximation point and the candidate point is greater than 1, and FIG. 6B shows the first order.
The order determination step is shown when the slope of the straight line connecting the approximate point and the candidate point is smaller than 1.

FIG. 7 is a detailed block diagram of the gradation / resolution information layering device 24. Gradation / resolution information layerer 24
Determines a pattern 75 for thinning out pixels on the basis of the order 28 from the important order determiner 23 (see FIG. 2), and a pixel thinning pattern generator 71 for outputting this, and thinning out pixels in the input block 9 according to the thinning pattern 75. And a pixel thinning-out signal generator 72 for outputting a pixel selection signal 77 for selecting a bit plane when the pixels in the input block 9 are divided into bit planes based on the order 28 and outputting the same. 73 and a gradation decimator 74 that selects a bit plane according to the gradation selection signal 77 and outputs the gradation resolution layered data 30.

The operation of the gradation / resolution information layering device 24 will be described below with reference to FIG. The order information 28 is sequentially output from the pixel thinning pattern generator 71 by the pixel thinning pattern generator 71.
Converted to 5. For example, the pixel thinning pattern 75 as shown in FIG. 8 is generated according to the pixel thinning rate corresponding to the resolution in the order information 28. FIG. 8 shows the correspondence between the pixel thinning rate and the pixel thinning pattern 75 when the size of the input block 7 is 8 × 8 pixels. The pixel thinning pattern shown in FIG. 8 is isotropic with respect to the two-dimensional direction of the image data, but for example, FIGS.
As shown in, the vertical and horizontal resolutions may be changed to form an anisotropic pixel thinning pattern. In this case, the pixel thinning pattern is adaptively changed according to the resolution analysis result 68, that is, the waveform shape of the input block 9. For example, in the input block 9 having edges in the vertical direction, high spatial frequency components are included in the horizontal direction, and high spatial frequency components are not included in the vertical direction, so that the original waveform shape is reproduced. As shown in FIG. 7A, the horizontal resolution may be high and the vertical resolution may be low. On the contrary, in the input block 9 having an edge in the horizontal direction, the pixel thinning pattern as shown in FIG.

The input block 9 is sequentially thinned by the pixel thinning unit 72 according to the pixel thinning pattern 75, and output as thinned pixels 76.

Further, the order information 28 is sequentially converted into the gradation selection signal 77 by the gradation selection signal generator 73. For example, according to the number of bits per pixel corresponding to the number of gradations in the order information 28, a gradation selection number 77 for selecting a bit plane as shown in FIG. 10 is generated. FIG. 10 shows the correspondence between the number of bits per pixel and the selected bit plane when the image data 6 has a gradation of 8 bits / pixel.

The pixels 76 from which the pixels have been thinned are sequentially thinned by the gradation thinning-out device 74 in accordance with the gradation selection signal 77 and output. As described above, the gradation / resolution information layering means 24 passes through the candidate points from the first approximation as shown in the example of FIG. Pixel patterns are sequentially output.

FIG. 11 is a detailed block diagram of the information source encoder 26. The information source encoder 26 is used by the importance ranking determiner 2
3 (see FIG. 2), a multiplexer 78 for multiplexing the gradation / resolution layered data 30 from the gradation / resolution information layerer 24, and an arithmetic code for the multiplexed data 80. If the target code amount reaching signal 29 from the code amount counter 25 (see FIG. 2) is not output, the code data 1
The arithmetic encoder 79 outputs 0 and terminates the encoding process when the signal 29 is output.

The information source encoder 26 will be described below with reference to FIG.
The operation of will be described. The order information 28 from the important order determiner 23 (see FIG. 2) and the gradation / resolution layered data 30 from the gradation / resolution information layerer 24 are candidates from the first approximation as in the example shown in FIG. Each time the hierarchical layer advances toward the original image point via the point, it is multiplexed by the multiplexer 78 and output. The information 80 multiplexed by the multiplexer 78 is
The coded data 10 is output after being encoded by the arithmetic encoder 79. In this embodiment, the arithmetic coding method is used for the source coding of the multiplexed information 80, but other methods such as the Huffman coding method may be used.

At this time, the rank information 28 includes information indicating that it is a candidate point (rank). Therefore, when the encoded layer is the layer estimated by the gradation / resolution candidate estimator 22 at the time of encoding by the arithmetic encoder 79, at the time of encoding the multiplexed rank information 28. , A code indicating a candidate point is generated and inserted.

Further, the code amount counter 25 of FIG. 2 adds up the total data amount of the code data 10, and when the total data amount reaches a predetermined code amount, or when the total data amount reaches the predetermined code amount. A target code amount reaching signal 29 for forcibly ending the encoding is output to the arithmetic encoder 79 of the information source encoder 26 immediately before reaching.

When the target code amount reaching signal 29 is input, the arithmetic encoder 79 ends the encoding and thereafter stops the output of the code data. Alternatively, the encoding may be completed immediately after inserting the identification code.

Next, another example of the configuration of the gradation / resolution information layering device 24 shown in FIG. 2 is shown in FIG. The gradation / resolution information layering device 24 shown in FIG. 12 is based on the JPEG method ("International standard for multimedia coding", edited by Yasuda, Maruzen, p.
(see p24-pp28), a hierarchical division method of progressive coding spectral selection (s-s) and successive aproximation (s-a) is used.

This spectral selection (s-s)
The scheme and the successful aproximation (sa) scheme will be briefly described with reference to FIG.

In FIG. 7A, the discrete cosine transform coefficient is 3
In the example shown in the figure, data in which the transform coefficient is expressed in 8 bits in the X axis direction, the dimension of the discrete cosine transform coefficient in the Y axis direction, and the block in the Z direction are taken.

In the spectral selection (s-s) method, as shown in FIG. 6B, scanning is started from the lower dimension side of the transform coefficient, and the first scan scans the 0th and 1st dimensions of the discrete cosine transform coefficient dimension. The next layer is selected and sequentially encoded in the block direction. Subsequently, the second and third hierarchical layers are selected in the second scan, and sequentially encoded in the block direction.
Thereafter, similar processing is repeated to encode all transform coefficients. That is, spectral selection (s-s)
In the method, encoding is performed from a high resolution image to a low resolution image.

Further, in the succeeding aproximation (s-a) system, as shown in FIG.
The scan is started from the MSB side of the transform coefficient, the 7th bit and the 6th bit of the transform coefficient are selected in the first scan, and sequentially encoded in the block direction. Then, in the second scan, the layers of the 5th bit and the 4th bit are selected and sequentially encoded in the block direction. Thereafter, similar processing is repeated to encode all transform coefficients. That is, in the succeeding aproximation (s-a) method, encoding is performed from an image with a sharp gradation change to an image with a gentle gradation.

The gradation / resolution information layering device 2 shown in FIG.
4 outputs a discrete cosine transform of the input block 9 and outputs a transform coefficient 86, and a transform coefficient selection signal 87 for selecting based on the rank 28 from the rank determiner 23 (see FIG. 2). Transform coefficient selection signal generator 8
2, a conversion coefficient selector 83 that selects from the conversion coefficient 86 according to the conversion coefficient selection signal 87, and a gradation selection signal generator 84 that outputs a gradation selection signal 89 that selects the bit plane of the conversion coefficient based on the order 28. , A gradation selector 85 that selects the bit plane of the conversion coefficient 88 selected according to the gradation selection signal 89 and outputs the gradation / resolution layered data 30.

The operation of the gradation / resolution information layering device 24 in the second embodiment will be described below with reference to FIG.

The rank information 28 is sequentially converted into a conversion coefficient selection signal 87 by the conversion coefficient selection signal generator 82, and the conversion coefficient 86 is converted by the conversion coefficient selector 83 accordingly.
Selected from. For example, as in the spectral selection (s-s) method shown in FIG. 13B, the transform coefficients are selected from the low dimension side to the high dimension side from the first scan to the nth scan.

The order information 28 is sequentially converted into a gradation selection signal 89 by the gradation selection signal generator 84, and the bit plane is selected and output by the gradation selector 85. For example, as in the successive aproximation (sa) method shown in FIG.
The scan coefficient is selected from the MSB side of the conversion coefficient toward the LSB side.

As described above, when the transform coefficient of the discrete cosine transform is selected on the basis of the order information 28, it is possible to perform the image coding with little deterioration of the decoded image quality with a small code amount. , ISO and CCITT
It becomes possible to perform encoding and decoding in conformity with the JPEG method, which is being standardized in the above.

In the principle configuration diagram shown in FIG. 1, the primary storage means 5 and the secondary storage means 7 are separately provided.
For example, the fixed-length encoded code data is temporarily stored in a secondary storage unit such as a magnetic disk, and the temporarily stored fixed-length code data is converted into variable-length code data and then stored in the same secondary storage unit. May be. Conversely, fixed length code data and variable length code data may be stored in the same primary storage means.

The following is a summary of the present invention described above.

In the present invention, an image is sampled, divided into blocks, the image quality of an input block coded / decoded is estimated, and the input block, the coded data, and the image quality estimation result are referred to for each one or a plurality of blocks. The coding parameters are determined so that the coding amount is less than or equal to a predetermined code amount, or less than or equal to a predetermined coding amount and less than or equal to a predetermined decoding image quality, and one or more input blocks are selected according to the coding parameters. Or less than or equal to or less than a predetermined code amount for each
Moreover, fixed-length (maximum code amount limitation) encoding is performed so that the image quality is equal to or lower than the predetermined decoding image quality. At this time, when the predetermined decoding image quality is reached, the identification code is inserted into the code data, and thereafter, the predetermined code amount continues Since the encoding process is terminated at the time of encoding or when the code amount necessary and sufficient for image quality reproduction is reached, the code data encoded with a certain code amount or less is temporarily stored in the primary storage means. , A constant compression rate is always obtained for each block, coding / decoding can be locally performed for each block, and uniform coding processing without complicated branch processing is performed. Further, since the identification code is detected from the temporarily stored code data and only the code data from the beginning of the code data of the block to the identification code is stored in the secondary storage means, sufficient decoding image quality is minimized. It can be obtained by the code amount.

[0070]

As described above, according to the present invention,
Providing means for estimating the decoded image quality when encoding the image,
Since the code amount is controlled according to the estimated image quality, it is possible to perform image coding such that a sufficient decoded image quality can be obtained with a small code amount.

[Brief description of drawings]

FIG. 1 is a block diagram showing a basic configuration of an image encoding device of the present invention.

FIG. 2 is a block diagram showing a configuration example of an image encoding device according to the present invention.

3 is a block diagram showing a configuration example of a gradation / resolution analyzer used in the image encoding device shown in FIG.

4 is a block diagram showing a configuration example of a resolution analyzer used in the gradation / resolution analyzer shown in FIG.

5 is a block diagram showing a configuration example of a gradation / resolution candidate estimator used in the image coding apparatus shown in FIG.

FIG. 6 is a diagram for explaining a procedure for determining the order of gradation candidates and resolution candidates in the order determining unit.

7 is a block diagram showing a configuration example of a gradation / resolution information layering device used in the image encoding device shown in FIG.

FIG. 8 is an explanatory diagram showing an example of a pixel thinning pattern in pixel thinning.

FIG. 9 is an explanatory diagram showing another example of a pixel thinning pattern.

FIG. 10 is a schematic diagram for explaining a bit plane selecting operation in gray scale thinning.

11 is a block diagram showing a configuration example of an information source encoder used in the image encoding device shown in FIG.

FIG. 12 is a block diagram showing another configuration example of the gradation / resolution information layering device.

FIG. 13 is an operation explanatory diagram of the second embodiment of the present invention.

FIG. 14 is a block diagram illustrating a configuration example of a conventional image encoding device.

[Fig. 15] Fig. 15 is a block diagram illustrating another configuration example of a conventional image encoding device.

FIG. 16 is a block diagram showing still another configuration example of a conventional image encoding device.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Blocking means, 2 ... Variable length encoding means, 3 ... Decoding image quality estimation means, 4 ... Code amount control means, 5 ... Primary storage means, 6 ... Variable length conversion means, 7 ... Secondary storage means, 8 ... Image data, 9 ... Input block, 10 ... Fixed length code data,
11 ... Decoded image quality estimation result, 12 ... Encoding parameter, 1
3 ... Temporarily stored fixed-length code data, 14 ... Variable-length code data, 18 ... Identification code detector, 19 ... Address control signal, 21 ... Gradation / resolution analyzer, 22 ... Gradation / resolution candidate estimator, Reference numeral 23 ... Important rank determiner, 24 ... Gradation / resolution information layerer, 25 ... Code amount counter, 26 ... Source encoder, 27 ... Gradation / resolution feature, 28 ... Rank, 29
... target code amount reaching signal, 30 ... gradation / resolution layered data, 31 ... gradation analyzer, 32 ... resolution analyzer, 33 ... multiplexer, 34 ... gradation analysis result, 35 ... resolution analysis result, 4
1 ... Block divider, 42 ... Average value separator, 43 ... First
Inner product calculator, 44 ... First vector set, 45 ... First index holder, 46 ... Second inner product calculator, 47 ... Second
Vector set, 48 ... Second index holder, 49
... third inner product calculator, 50 ... third vector set, 51 ...
Third index holder, 52 ... Divided block, 53 ...
Average value separation block, 54 ... First representative vector, 55 ...
Code of first inner product result, 56 ... First index, 57 ...
Second representative vector, 58 ... Sign of second inner product result, 59 ...
Second index, 60 ... Third representative vector, 61 ... Sign of third inner product result, 62 ... Distributor, 63 ... Gradation candidate estimator, 64 ... Resolution candidate estimator, 65 ... Multiplexer, 66 ... Gradation feature , 67 ... Gradation candidate, 68 ... Resolution feature, 69 ... Resolution candidate, 71 ... Pixel thinning pattern generator, 72 ... Pixel thinning pattern, 73 ... Gradation selection signal generator, 74 ... Gradation thinning pattern, 75 ... Pixel Thinning pattern, 76 ... Block after pixel thinning, 77 ... Gradation selection signal, 78 ... Multiplexer, 79 ...
Arithmetic encoder, 80 ... Encoding target data, 81 ... Discrete cosine transformer, 82 ... Transform coefficient selection signal generator, 83 ...
Conversion coefficient selector, 84 ... gradation selection signal generator, 85 ... gradation selector, 86 ... conversion coefficient, 87 ... conversion coefficient selection signal,
88 ... Selected conversion coefficient, 89 ... Gradation selection signal

Claims (7)

[Claims] 1. An image sampled from a plurality of pixels, m ×
Blocking means for dividing the input block into n pixels (m and n are positive integers), coding means for coding the input block, and code quantity for controlling the code quantity at the time of coding in the coding means. An image coding apparatus comprising a control means, an image quality estimating means for estimating the image quality when the input block is coded and decoded is provided, and the coding means estimates the image quality for each predetermined image area by the image quality estimating means. The encoded image quality is referred to so as to gradually increase from the decoded image quality to a higher decoded image quality, and when the predetermined decoded image quality is reached, the identification code is inserted into the code data, and the code is encoded so that the predetermined code amount is obtained. An image encoding device characterized by encoding. 2. An image of an image of m × consisting of a plurality of pixels is sampled.
Blocking means for dividing the input block into n pixels (m and n are positive integers), coding means for coding the input block, and code quantity for controlling the code quantity at the time of coding in the coding means. In an image coding device comprising a control means, an image quality estimation means for estimating the image quality when the input block is coded and decoded is provided, and in the coding means, for each predetermined image area, a predetermined code amount or less An image coding apparatus is characterized in that coding is performed so as to have a predetermined decoded image quality or less, and an identification code indicating that the coding is completed is added to the coded data. 3. The decoded image quality estimating means has an analyzing means for analyzing the resolution and gradation feature amount of the pixels in the input block, and estimates the decoded image quality based on the analysis result by the analyzing means. The image coding apparatus according to claim 1 or 2, wherein 4. An analysis unit for analyzing the feature quantity of the resolution determines m × n pixels (m and n are positive integers) obtained in advance, or pixels divided by the positive integer ratio j (j is a positive integer). And the average value separation block obtained by subtracting the average value in the input block from each pixel in the input block, the degree of approximation is determined, and the highest degree of approximation is obtained. 4. The image code according to claim 3, wherein the index of the representative shape block or a set of indexes of the representative shape block having the highest degree of approximation for each of the j divided blocks is used as the feature amount of the resolution of the input block. Device. 5. The code amount control means has an important order determining means, and the important order determining means sets a hierarchy of a predetermined resolution and a hierarchy of gradations as a first order, and the decoded image quality estimating means. As a candidate, a layer of resolution and a gradation having an image quality estimated by the above are used as a candidate of the layer of resolution and a candidate of the layer of gradation starting from the layer of resolution and the layer of gradation of the first rank. The image coding apparatus according to claim 1 or 2, wherein the image coding apparatus ranks toward each other. 6. The encoding means has a layer dividing means for dividing the input block into layers of gradation and resolution, and the layer amount dividing means controls the code amount of each layer for each predetermined image area. The image encoding device of the image encoding device according to claim 1 or 2, wherein the image encoding device sequentially performs the encoding in accordance with the order determined by the means. 7. An image coding apparatus for converting code data coded by the image coding apparatus according to claim 1 or 2, further comprising means for identifying the identification code. An image coding apparatus, wherein the identification code identifying means converts the code data by selecting only the code data from the beginning of the code data to the identification code for each predetermined image area.