JP3330641B2 - Recording transmission coding method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image data compression encoding system, and more particularly to an encoding system for efficiently recording and transmitting still images.

[0002]

2. Description of the Related Art At present, the most general high-efficiency coding method for image data is shown in FIG. The input image data 11 is compressed by the encoding unit 12 to become the encoded data 14. At this time, the compression is basically controlled by an external setting parameter so that the image quality of the decoded image is constant. Here, this parameter will be referred to as an image quality control parameter. FIG.
Corresponds to the image quality control parameter 13. As a result, it is possible to always keep the image quality constant and compress the image regardless of the complexity of the image to be handled. However, the code data amount changes depending on the redundancy of the image. That is, an image with a high degree of redundancy has a small amount of code data, but an image with a low degree of redundancy has a large amount of code data.

For example, JPEG (Joint Photographic C), which is a standard method of still image coding, is used.
odding Express Group) is also an encoding method included in FIG. (For details, see ISO / IEC
JTC1 / SC2 / WG10: JPEG Techn
Ial Specification, JPEG8-
See R8 (1990.8). The image quality control parameter 13 in this JPEG corresponds to a scaling factor for weighting quantization. Also, Japanese Patent Application Laid-Open No. 62-252
The non-equal-length block division coding method described in No. 217 “Data coding method” is also a coding method included in FIG. The image quality control parameter 13 in this encoding method is
This corresponds to an allowable value for determining the magnitude of distortion, that is, allowable distortion.

By the way, a moving picture is compressed and digital V
Considering the case of recording or transmitting in a TR, it is inconvenient for the amount of code data to vary depending on the redundancy of the image as described above. In this case, for the encoding device, since the input data rate r (in) and the output data rate r (out) are fixed, the compression ratio is always r regardless of the redundancy of the image.
(Out) / r (in). Therefore, it is necessary to determine an image quality control parameter for obtaining a given amount of code data according to the redundancy of the image. This control is herein referred to as code data amount control.

There are several techniques for controlling the amount of encoded data in the encoding method shown in FIG. An example is shown in FIG.
Shown in The image quality control parameter determination unit 18 in FIG. 3 determines the redundancy of the input image 15 from the redundancy data 17, and considers the redundancy of the image in order to obtain the code data amount 19 provided from the outside. Appropriate image quality control parameters 20
Is output. This output image quality control parameter 20
If the encoding is performed by the encoding unit 16 based on the above, the encoded data 21 of the given amount of encoded data can be obtained.

When the non-equal-length block division coding method enables the control of the code data amount, the redundancy data includes distortion (original image) when all the possible blocks are approximated by a function in each block. And the difference). When this value is large as a whole, the redundancy is small, and when this value is small as a whole, the redundancy is large.

The usefulness of this code data amount control is not limited to moving images. For example, when a still image is compressed and recorded on a disk or the like, since the capacity of the recording medium is determined, if the amount of code data can be controlled, the number of recorded images can be completely guaranteed. This is essential for an electronic still camera. Also, in the transmission, when a still image is transmitted from the site for live broadcast news or the like, the still image must be transmitted within a limited time. This is also made possible by controlling the amount of code data. As described above, the control of the code data amount is a sufficiently necessary function even for a still image, and there are actually many systems.

However, when code data amount control is applied to a still image, it is not always efficient. This is because the images to be handled include images with various degrees of redundancy. In the case of normal code data amount control, a larger code data amount is set so that even if an image having less redundancy is compressed, a problem does not occur in image quality deterioration. Therefore, the code data amount set for most images is too large. For an image having more redundancy, it is sufficiently conceivable that no problem occurs in the image quality of the decoded image even if the data amount is 1/2, 1/4, or less than the set code data amount.

[0009]

As described above, by using the conventional code data amount control, it is possible to guarantee the number of recordings and perform the compression recording of a still image and the compression transmission within a limited period. On the other hand, the redundancy has not been sufficiently reduced for an image having more redundancy, and efficient compression has not been realized. In order to solve these drawbacks, the present invention does not exceed the set code data amount and, for an image with much redundancy, to the extent that the image quality limited by the externally set image quality control parameter is met. An object of the present invention is to enable efficient coding by further reducing the amount of code data.

[0010]

FIG. 1 is a block diagram showing the overall configuration of the present invention. In FIG. 1, an input image 1 is input to an encoding unit 2, and redundancy data 3 is output from the encoding unit 2 and input to an image quality control parameter determination unit 4. This image quality control parameter determination unit 4 stores a code data amount 5
, And outputs the image quality control parameter 6 as an output. The image quality control parameter 6 is input to the image quality control parameter determination unit 8, and the reference image quality control parameter 7 is also input to the image quality control parameter determination unit 8. The image quality control parameter determining unit 8 outputs the optimal image quality control parameter 9 to the encoding unit 2. The encoding unit 2 outputs code data 10.

[0011]

The operation of the present invention will be described below. 1 is the same as the conventional one until the image quality control parameter determination unit 4 outputs the image quality control parameter 6. That is, the encoding unit 2 receives the input image 1 and outputs redundancy data 3 to the image quality control parameter determination unit 4 as information indicating the redundancy of the image. The image quality control parameter determination unit 4 outputs an image quality control parameter 6 for compressing the code data amount 5 to an externally set code data amount.

The operation of the image quality control parameter determining section 4 will be described more specifically with reference to FIG. First, the horizontal axis in FIG. 4 indicates the code data amount set externally. The vertical axis indicates the value of the image quality control parameter 6. (It is assumed that the larger the value is, the greater the deterioration of the image quality of the decoded image is.) The lines a, b, and c indicate the redundancy of each image obtained from the redundancy data 3. A line closer to the origin indicates an image with more redundancy. It is assumed that the redundancy of the image handled here is an image in the range of approximately a to b. If the value of the image quality control parameter in which the image quality deterioration does not matter is supposed to be 16,
In an image having less redundancy (image of the line b), the image quality control parameter becomes 16 when the code data amount is 400K.
Byte. Therefore, the code data amount 5
Is set to 400 KByte.

The operation of the image quality control parameter determining unit 4 is as follows. If an image having the redundancy a is input, it is determined from the redundancy data 3 that the redundancy is the redundancy of the line a shown in FIG. Then, the value of the image quality control parameter at the intersection of the value 400 KByte set to the code data amount 5 and the line a is obtained. In this case, since it is 8, this 8 is output as the value of the image quality control parameter 6.
In some cases, an image having less redundancy than the assumed redundancy of b may be input. Also in this case, the image quality control parameter determination unit 4 sends the redundancy data 3
From this, for example, it is understood that the redundancy of the line c,
The image quality control parameter value 32, which is the intersection of the line and the code data amount 5 value 400 KByte, is set to the image quality control parameter 6
Output as Thus, the image quality control parameter determining unit 4
Is to output an image quality control parameter 6 in which the code data amount becomes 400 KBytes regardless of the image of any redundancy.

What is problematic in the present invention is that if the frequency of an image close to the redundancy of the line a is high in the image to be handled,
This is an inefficiency due to the fact that most images are compressed with image quality control parameter values much lower than 16 where image quality degradation problems do not occur. In other words, there is an inefficiency in that the amount of code data becomes more than necessary in order to perform compression with higher image quality than necessary. And this is a common case in reality.

Therefore, in FIG.
Is not directly input to the encoding unit 2, but is compared once with the reference image quality control parameter 7 by the image quality control parameter determination unit 8, changed to an appropriate value, and input to the encoding unit 2. The value of the reference image quality control parameter 7 is set to a value that gives an image quality that does not pose a problem in practical use. In the example of FIG.
The value is 6. The specific processing of the image quality control parameter determination unit 8 is to compare the image quality control parameter 6 with the reference image quality control parameter 7 and determine a larger value as the optimum image quality control parameter 9. That is, optimal image quality control parameter 9 = MAX (image quality control parameter 6, reference image quality control parameter 7)

As a result, an image having a high degree of redundancy a is input, and the value of the image quality control parameter 6 is set to 8 (8 <8 <
Even in the case of <16), the image quality control parameter determination unit 8 can change the optimum image quality control parameter 9 to a value of 16 and output the value to the encoding unit 2. As a result, the image can be compressed to a useless image quality, and the code data amount is also reduced to 20 as shown in FIG.
It is possible to encode a data amount of 0 KByte, which is sufficiently smaller than 400 KByte. The image quality control parameter determination unit 8 determines whether an image having less redundancy than the assumed one is input, for example, if an image having the redundancy of the line c in FIG. 6 becomes 32 (32 >> 16), and the optimum image quality parameter 9
Is output as 32. For this reason, it is possible to perform encoding without exceeding the code data amount as in the past.

[0017]

FIG. 5 shows an embodiment of the present invention. The input image 22 is input to the unequal-length block division encoding unit 2.
The distortion data 24 is output from the unequal-length block division encoding unit 23 to the allowable distortion determination unit 25. The allowable distortion determination unit 25 also receives the code data amount 27 and outputs an allowable distortion 26. The output allowable distortion 26 is input to the selector 28 and the comparator 29. Also, the selector 28
The comparator 29 also receives the reference allowable distortion 31. The comparator 29 outputs a selector control signal 30 to the selector 28. The optimum allowable distortion 32 is output from the selector 28 to the unequal-length block division encoding unit 23. Then, the coded data 33 is output from the unequal-length block division coding unit 23.

The unequal-length block division encoding unit 23 comprises:
Basically, processing is performed by the coding method of Japanese Patent Application Laid-Open No. 62-252217, "Data Coding Method". The output distortion data 24 corresponds to the distortion (difference from the original image) when function approximation is performed on all blocks that can appear by the above-described encoding method. The input optimum allowable distortion 32 corresponds to an allowable value for determining the magnitude of the distortion. Therefore, the operation of the unequal-length block division encoding unit 23 encodes the input image 22 so as not to deteriorate from the image quality limited by the given optimum allowable distortion 32. The distortion data 24 calculated during the processing is output. The allowable distortion determining unit 25 is a processing unit that enables control of the amount of code data. The basic processing content is to determine the redundancy of the image from the distortion data 24, and to set the code data amount 2
7 to obtain an allowable distortion 26 for compression to 7.

FIG. 6 shows a more specific embodiment of the allowable distortion determining section 25. The histogram creation unit 35 of FIG. 6 receives the distortion data 34 of all the possible blocks, creates a histogram with the distortion value on the horizontal axis, and stores the histogram in the storage unit 37 by the histogram writing data 36. . Next, the allowable distortion detection unit 39 first obtains a code data amount when the allowable distortion is set to 、,
This is defined as the initial code data amount. When the allowable distortion is set to ∞, the target data is always divided only by the block of the maximum size regardless of the content of the input data, so the code data amounts allocated to these blocks are added. It can be obtained only by: This initial code data amount is the smallest data amount, and the allowable distortion is reduced.
As the size becomes smaller, the amount of code data increases. Therefore, the allowable distortion detection unit 39 obtains the amount of code data that increases when the allowable distortion is set to a value smaller than 次 に by inputting and examining the histogram read data 38 stored in the storage unit 37.

Specifically, if the allowable distortion is set to a value smaller than 例 え ば, for example, 254, the number of blocks having the distortion of 254 or more is obtained from the histogram. Blocks having these allowable distortions or more are replaced with four child blocks corresponding to the corresponding block in the quadtree structure to obtain an allowable distortion value of 2
The state changes to a division state satisfying 55. In other words, since all blocks that can appear have a relationship that the distortion of the child block in the quadtree structure is not greater than the distortion of the parent block, blocks that are higher than the allowable distortion are replaced with the child blocks to reduce the allowable distortion. That is, the state changes to the satisfied block division state. Although this change is referred to as block expansion here, when one block is expanded as described above, it is replaced with four child blocks, so that it is expanded from the amount of code data allocated to all four child blocks. By subtracting the amount of code data allocated to a given block, it is possible to obtain an increased amount of code data by one block expansion. Assuming that this data amount is 6 bytes and the number of blocks (blocks having a distortion of 255 or more) to be developed from the histogram when the allowable distortion is 255 as described above is 100, , The amount of increase of the code data is obtained by 6 × 100. If the increase due to the one-block expansion is not limited to 6 bytes but is of a plurality of types, if the types of blocks that can appear are classified for each type and a histogram is created for each of the classified blocks, code data is similarly obtained. Can be obtained. By adding the increase amount obtained as described above to the initial code data amount, the code data amount for the reduced allowable distortion is obtained. Then, the above processing is repeated while changing the allowable distortion so that the code data amount becomes equal to the code data amount 40 specified from the outside. As a result, the allowable distortion detector 39 can obtain an allowable distortion that can be encoded into the code data amount 40, and can obtain it as the allowable distortion 41.

The image quality control parameter judging section 8 of FIG. 1 which is a feature of the present invention is realized by a selector 28 and a comparator 29 in the embodiment of FIG. The comparator 29 determines the magnitude relation between the allowable distortion 26 and the reference allowable distortion 31. If the allowable distortion 26 is large, the selector control signal 30 controls the selector 2 so that the allowable distortion 26 is connected to the optimum allowable distortion 32.
8 is controlled. When the allowable distortion 26 is small,
The selector 28 is controlled by the selector control signal 30 so that the reference allowable distortion 31 is connected to the optimum allowable distortion 32.
(If the values are exactly the same, it does not matter which one is connected.)

As a result, when an image having a higher degree of redundancy is input, the allowable distortion 26 has a value smaller than the reference allowable distortion 31. The comparator 29 determines this, and controls the selector 28 based on the selector control signal 28 so that the optimum allowable distortion 32 becomes the reference allowable distortion 31. Therefore, it is possible to efficiently encode an image having a higher degree of redundancy with a smaller amount than the value set by the code data amount 27 without compressing the image with a higher image quality than necessary. . Also,
When an image having less redundancy than the assumed redundancy is input, the allowable distortion 26 becomes a value larger than the reference allowable distortion 31. The comparator 29 determines this, and selects the selector control signal 28
Thus, the selector 28 is controlled so that the optimum allowable distortion 32 becomes the allowable distortion 26. As a result, encoding can be performed without exceeding the value set in the code data amount 27 as in the related art. If this is applied to a compression recording transmission system, it is possible to record more images than the guaranteed number of recordings or to transmit in a shorter time than the limited time according to the redundancy of the image. And an efficient recording and transmission system can be realized. The processing in the comparator 29 and the selector 28 in FIG.
It is also possible to realize by software using cpu. A flowchart at this time is shown in FIG.

[0023]

According to the present invention, an image quality control parameter for compressing to a predetermined code data amount and a reference image quality control parameter for achieving a predetermined image quality are compared and determined. As a result, it is possible to reduce the amount of code data for any image as before, and for images with more redundancy, to reduce the amount of data with a smaller amount of data without deteriorating the image quality than the standard image quality control parameter. Good encoding is possible.

[Brief description of the drawings]

FIG. 1 is a block diagram showing the overall configuration of the present invention.

FIG. 2 is a block diagram illustrating an outline of a general high-efficiency coding method for an image.

FIG. 3 is a block diagram showing an outline of a system in which the amount of code data is controllable in the related art.

FIG. 4 is a diagram for explaining the operation of an image quality control parameter determining unit 4;

FIG. 5 is a diagram showing one embodiment of the present invention.

6 is a block diagram of a more specific embodiment of the allowable distortion determining unit 25 in FIG.

FIG. 7 is a diagram showing an example of a flowchart in a case where the processing of the comparator 29 and the selector 28 of FIG. 5 is realized by software.

[Explanation of symbols]

 6 Image quality control parameters 7 Reference image quality control parameters 8 Image quality control parameter determination unit 8 9 Optimal image quality control parameters

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int.Cl. ⁷ , DB name) H04N ^7/ 24-7/68 H04N 1/41-1/419 H04N 5/91-5/956

Claims (2)

(57) [Claims] An encoding method for encoding by controlling an image quality control parameter value corresponding to a scaling factor for weighting quantization so that input image data is encoded to a predetermined code data amount. when encoding image data, and the image quality control parameter values at that time, an image quality control parameter values when the image quality of the image data is encoded in a predetermined image quality allowable preset reference quality control parameter values compare, select the larger parameter values of the parameter values as an optimal image quality control parameter values, it is encoded with the image quality control parameter values, and more
A recording and transmission coding method characterized by efficiently coding with a small amount of code data . 2. A distortion which is a difference between the input image and the decoded image.
Image quality control parameters calculated based on the size ofvalue
And the preset reference image quality control parametervalueCompare
And the larger of the parameter valueParameter valueThe best image quality
Control parametersvalueSelected asThe image quality control parameter value
By encodingEfficient with less code data
2. The recording transmission according to claim 1, wherein the data is encoded.
Encoding method.