JPH07123270A - Picture coding and decoding processing method - Google Patents

Abstract

PURPOSE:To attain synchronization between parallel elements at coding and decoding by dividing picture information so as to have similar statistic property before coding and dividing the information to be a state before coding synthesis. CONSTITUTION:Inputted picture information at coding is subject to division processing so that statistic properties of divided pictures are close to each other at a picture information division synthesis means 101 and the divided picture information is inputted to plural coding decoding means 102, 102',... 102'' to attain parallel coding processing. On the other hand, in the case of decoding, the received code information is divided into a state before synthesis at coding by a code information division synthesis means 103 and the divided code information is inputted to the plural coding decoding means 102, 102',... 102'', in which parallel decode processing is conducted. Through the constitution above, synchronization between parallel elements is attained even at coding and decoding and a deviation of a timing for the parallel processing is reduced to enhance the effect of processing time reduction by the parallel processing.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method of dividing image information and performing coding / decoding processing in parallel.

[0002]

2. Description of the Related Art When a large amount of digital image data is handled by a computer or the like, it is necessary to introduce an image coding method that achieves high compression (hereinafter referred to as a high compression coding method).
Therefore, many encoding methods have been proposed in the past, and JPEG (Joint Photographic) has been proposed.
Expert Group) and JBIG (Joint)
Bi-level Coding Expert Gr
For example, an encoding method which is an international standard has been studied / standardized.

FIG. 12 shows a baseline system (Baseline System) which is an essential function of the JPEG system.
The encoder and the decoder of m) are illustrated. The encoder is a DC
T (Discrete Cosine Transfer)
rm: Discrete Cosine Transform) section 1001 and quantization section 10
02 and an entropy encoding unit 1003, and the decoder includes an entropy decoding unit 1004 and an inverse quantization unit 10
05 and an IDCT (Inverse DCT) unit.

[0004]

While the encoding system is required to achieve high compression, it is also required that the decoded image has high image quality. Therefore, JPE
As shown in the example of the G method, the encoding / decoding algorithm is complicated and requires a long processing time. The processing time required for this encoding / decoding becomes a problem particularly when a high-definition image having a large number of pixels is processed at high speed.

The following three methods can be mainly considered as means for shortening the processing time.

1) For example, the processing speed of hardware is increased by increasing the clock frequency.

2) Simplify the processing.

3) Divide an image and process it in parallel.

However, these means have the following problems.

The speeding up of the processing of 1) depends on the characteristics of the device and has a limit.

The simplification of the process 2) makes it difficult to achieve both high compression and high image quality.

In parallelization of the process 3), it is difficult to synchronize the parallel elements.

Here, the problem of parallelization of the processing of 3) will be described. FIG. 13 illustrates a schematic diagram of an encoding / decoding processing device that divides an image into a plurality of regions and processes the images in parallel. 1
Reference numeral 101 is an image information area division / synthesis unit, 1102 and 110.
Reference numerals 2 ′ and 1102 ″ are an encoding / decoding unit, and 1103 is a code information division / synthesis unit.

At the time of encoding, the input image information is divided into a plurality of areas by the image information area division / combination unit 1101 and input in parallel to a plurality of encoding / decoding units 1102, and each stage of the encoding / decoding unit 1102. The code information output from the above is combined in the code information division combining unit 1103.

At the time of decoding, the input code information is divided by the code information division / combination unit 1103, is input in parallel to a plurality of encoding / decoding units 1102, and is a decoded image output from each stage of the encoding / decoding unit 1102. The information is composed in the same way as when it was input.

In FIG. 13, the code information division / synthesis unit 11 is used.
03 or the image information division / synthesis unit 1101 does not include a large-capacity storage device, the processing time Ti of the entire device
me is proportional to Number / CR / Rate. However, Number is the number of pixels for encoding / decoding processing, CR
Is a compression rate, and Rate is a read speed from the code information division / combination unit 1103 or the image information division / combination unit 1101.

Generally, since the read rate Rate is constant, it can be seen that the processing time Time changes according to the variation of the compression rate. In order to enable synchronous control of parallel encoding / decoding processing, it is necessary to keep the compression ratio of each divided image within a certain range.

Generally, in the coding method, it is possible to control the degree of compression, that is, whether to perform high compression or to give priority to image quality with low compression, to some extent by adjusting the coding parameter. However, in the high compression coding method, the compression rate strongly depends on the statistical property of the input image, and therefore the code amount of the result, that is, the compression rate to be achieved cannot be known unless the coding is performed. Therefore, as a code amount control method for achieving the target compression rate, the following 2
One is often used.

A) Feedback control b) Feedforward control utilizing statistical properties of control target image information In the feedback control of a), the control target image information is encoded once, and the resulting code amount and target code amount are Compare
This is a method of adjusting the coding parameters so that the target code amount is achieved, and then performing the coding again to control the code amount. This method causes a complicated configuration and delays in control, and thus is not applied when aiming at high-speed processing.

In the feed-forward control of b), the statistical property of the control target image information is examined before the coding, the coding is performed using the coding parameter according to the statistical property, and the target compression rate is obtained. Is a control method that achieves This means that when image information having similar statistical properties is encoded using the same encoding parameter, it can be expected to be encoded at a substantially constant compression rate.

In the example of FIG. 13, for example, if the image information is simply divided into regions for the number of parallel stages and the encoding processing is performed in parallel, the statistical properties of the individual regions do not necessarily become the same. Therefore, it is considered that the compression rate does not become constant and synchronous control becomes difficult.

The present invention has been made to solve the above problems, and an object thereof is to divide an image and process them in parallel in order to speed up the encoding / decoding process. , Before encoding, by dividing the image information so that the statistical properties are close to each other, the compression ratios of the divided images in each coding unit are autonomously made approximately equal, and the coding is performed at the time of decoding. (EN) An image encoding / decoding processing method that enables synchronization between parallel elements at the time of encoding and decoding by dividing code information so that it is parallelized as in the case of time. Especially.

[0023]

In order to achieve the above object, the present invention provides an image information division / combination means for dividing or combining image information and a plurality of encoding / decoding means for encoding or decoding image information in parallel. And an image coding / decoding processing method in an image coding / decoding processing including a code information dividing / combining unit that divides or combines code information.
The input image information is divided by the image information division / combination means so that the statistical properties of the divided image information are close to each other, and the divided image information is input to each of the plurality of encoding / decoding means and encoded in parallel. And the plurality of code information output from each of the plurality of encoding / decoding means are combined in the code information division / combination means into combined code information, and the input combined code information is divided into the code information division at the time of decoding. The combining means divides the divided code information into a plurality of pieces of divided code information, inputs the plurality of pieces of divided code information to each of the plurality of encoding / decoding means, decodes in parallel, and outputs from each of the plurality of encoding / decoding means. It is characterized in that a plurality of pieces of decoded image information are combined by the image information division combining means to form combined image information.

[0024]

The operation of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a basic configuration diagram of an image coding / decoding processing apparatus according to the present invention. As shown in FIG. 1, an image coding / decoding processing apparatus of the present invention includes an image information division / combination means 101 for dividing or combining image information, and a plurality of encoding / decoding means 102 for encoding or decoding image information in parallel. , 1
02 ', ..., 102 "and code information division / combination means 103 for dividing or combining code information.

In the above configuration, at the time of encoding, the input image information is divided into a plurality of image information by the image information division / combination means 101, and the divided image information is divided into a plurality of encoding / decoding means 102, 102 ',. .., 102 ", and performs parallel encoding processing.
In No. 1, since the division processing is performed so that the statistical properties of the divided images are close to each other, the encoding / decoding means 102, 102 ′,
······················ is substantially equal to each other, the encoding and decoding means 102, 102 ′ ,.
, 102 "from each of the code information division / synthesis means 10
There is little deviation in the output timing to 3, and there is almost no useless waiting time until composition.

At the time of decoding, the input code information is divided into a plurality of code information by the code information division / combination means 103,
This division code information is converted into a plurality of encoding / decoding means 102, 10.
2 ', ..., 102 "are input to perform decoding processing in parallel. In the code information division / combination means 103, since the code information is divided into the state before the combination at the time of encoding,
The divided code information can be decoded in almost the same processing time, and the coding / decoding means 102, 102 ', ..., 102 "
There is little deviation in the output timing of the image information from each of the above to the image information division / combination means 101, and a useless waiting time before the composition hardly occurs.

Therefore, at the time of both encoding and decoding, it is possible to perform the processing efficiently by utilizing the effect of shortening the processing time due to the parallelization.

[0029]

FIG. 2 is a diagram showing a configuration of an image coding / decoding processing apparatus according to a first embodiment of the present invention, and FIG. 3 shows a signal flow when encoding is performed by the apparatus shown in FIG. FIG. 4 is a diagram showing a signal flow when decoding is performed by the apparatus shown in FIG.

In the figure, 1 is input image information, 2 is a scanning order conversion unit for converting the scanning order between raster scanning and block scanning, and 3 is a scanning order conversion unit 2 which blocks input image information 1 in the raster scanning order. Scanning order conversion image information converted into scanning, 4 is an image information storage unit that stores image information in a block scanning order, 5 is blocked image information, 6
Is an image information division / synthesis unit that divides image information or combines the divided elements so that the correlation of each element is high,
7 ', ..., 7 "are divided block image information obtained by dividing the block image information 5 in the image information division / combination unit 6,
8, 8 ', ..., 8 "are first to n-th encoding / decoding units for performing encoding or decoding processing in parallel, 9, 9', ..., 9"
, Is divided code information output from the encoding / decoding units 8, 8 ′, ..., 8 ″, 10 is a code information division / synthesis unit that synthesizes or divides the code information, 11 is synthetic code information, 12, 12 ′,
.., 12 ″ is the decoded divided blocked image information output from the encoding / decoding unit 8, 13 is the decoded divided blocked image information 12, 12 ′, ...
The decoded block image information synthesized in (1), 14 is the decoded raster scan order image information read from the scan order conversion unit 2 in the raster scan order, and 15 is the decoded image information.

FIG. 5 shows, in the above-mentioned first embodiment,
16-pixel × 16-line blocked image information 505
In the vertical and horizontal directions, every other pixel is selected to select four divided block image information 501 of 8 pixels × 8 lines.
An example of division in the case of division into 504 is shown. For example, it is known that the correlation between adjacent pixels in an image is generally high, as shown in Takahiko Fukibuki, "Digital Signal Processing of Image", published by Nikkan Kogyo Shimbun.
Therefore, if division is performed as shown in FIG. 5, divided block image information 501 to 504 having similar statistical properties can be obtained.

In FIG. 6, division code information 601 to 604, which is the result of encoding the block image information 501 to 504 divided as shown in FIG. 5 in the first embodiment, is synthesized as synthesis code information 605. An example is shown.

Next, the operation will be described. At the time of encoding, the input image information 1 in FIG.
In block scanning order, the converted block scanning order image information 3 is stored in the image information storage unit 4. The blocked image information 5 read from the image information storage unit 4 is divided by the image information division / synthesis unit 6 into divided block image information 7, 7 ′, ..., Which have similar statistical properties as in the division example of FIG. 7 ", each of which is encoded by a plurality of encoding / decoding units 8, 8 ', ..., 8",
The divided code information 9, 9 ', ..., 9 "output from each of the encoding / decoding units 8, 8', ..., 8" is combined by the code information division combining unit 10 as shown in FIG. .

At the time of decoding, in FIG. 4, the combined code information 11 is divided by the code information dividing / combining unit 10 into the state before combining at the time of encoding as shown in FIG. 6, and a plurality of encoding / decoding units 8, 8 ', .., 8 ", and a plurality of pieces of decoding division block information 12, 12 ',.
Are combined into a state before encoding as shown in FIG. 5 in the image information division combining unit 6, and the decoded block image information 13
Is stored in the image information storage unit 4. The scanning order conversion unit 2 converts the decoded blocked image information 13 stored in the image information storage unit 4 into a raster scanning order, reads it as the decoded raster scanning order image information 14, and decodes the decoded image information 15
Is output.

FIG. 7 shows the configuration of the image coding / decoding processing apparatus according to the second embodiment, and FIG. 8 shows the configuration of the image coding / decoding processing apparatus according to the third embodiment. The parts corresponding to those of the first embodiment shown in FIG. 2 are designated by the same reference numerals and the description thereof will be omitted. In the figure, 16 is a decoded reduced image, 17 is a reduced image synthesizing unit, and 18 is synthetic decoded reduced image information.

In the example of FIG. 7, the first encoding / decoding unit 8 of the encoding / decoding units 8 is controlled by the image information division / synthesis unit 6.
The decoded divided blocking information output from the image information storage unit 4 is stored, and the decoded raster scanning order image information from the image information storage unit 4 is read by the scanning order conversion unit 2 in the raster scanning order to obtain the original image. The 1 / n decoded reduced image 16 is obtained.

Further, in the example of FIG. 8, the decoding division blocking information output from the first encoding / decoding unit 8 of the encoding / decoding unit 8 is controlled by the image information division / synthesis unit 6 so that the image information storage unit 4 stores the decoding division blocking information. The decoded raster scan order image information from the image information storage unit 4 is read by the scan order conversion unit 2 in raster scan order to obtain a 1 / n decoded reduced image 16 of the original image, and further reduced. The image combining unit 17 combines n pages of the decoded reduced image 16 to obtain combined decoded reduced image information 18. FIG. 9 shows a synthesis example of the decoded reduced image information in the case of n = 4. In FIG. 9, the result of combining the quarter reduced image information 901 ′ to 904 ′ of the image information 901 to 904 for four pages is the combined decoded reduced image information indicated by reference numeral 905.

In the above-described embodiment, the divided block image information is obtained by selecting every other pixel in the vertical and horizontal directions of the pixels of the original image so that the statistical properties of the divided image information are close to each other. However, the present invention is not limited to this. For example, as shown in FIG. 10, the block image information 1201 of 24 pixels × 16 lines is selected every other pixel in the vertical direction and every two pixels in the horizontal direction, and six divided blocks of 8 pixels × 8 lines are selected. The divided image information 1202 to 1207 is obtained, or as shown in FIG. 11, the blocked pixel information 1301 of 24 pixels × 24 lines is selected every two pixels both vertically and horizontally to obtain nine divided block image information. It is also possible to obtain 1302 to 1313.

[0039]

As described above, according to the present invention, when the processing time is shortened by parallelizing the encoding and decoding, the image information or the code information between the parallel elements, that is, the object of the parallel processing is obtained. The division is performed so that the statistical properties between them are close to each other. As a result, it is possible to reduce the deviation of the timing of the parallel processing and enhance the effect of shortening the processing time due to the parallelization.

Further, by using only the decoded image output from one encoding / decoding unit at the time of decoding, a reduced image of 1 / n of the original image can be approximated in a decoding time of 1 / n. Moreover, it is possible to obtain the data without performing the reduction process again.

By combining n pages of the reduced image of 1 / n, an image obtained by combining reduced images of a plurality of pages into one page can be decoded in about 1 / n, and
It is possible to obtain the data without performing reduction processing again.

[Brief description of drawings]

FIG. 1 is a basic configuration diagram of an image encoding / decoding processing device according to the present invention.

FIG. 2 is a schematic configuration diagram of an image coding / decoding processing apparatus according to a first embodiment of the present invention.

FIG. 3 is a block diagram showing a signal flow at the time of encoding in the embodiment shown in FIG.

FIG. 4 is a block diagram showing a signal flow at the time of decoding in the embodiment shown in FIG.

FIG. 5: Division of image information in the embodiment shown in FIG.
It is a figure showing an example of composition.

FIG. 6 is a diagram showing the division of code information in the embodiment shown in FIG.
It is a figure showing an example of composition.

FIG. 7 is a schematic configuration diagram of an image encoding / decoding processing device according to a second embodiment.

FIG. 8 is a schematic configuration diagram of an image encoding / decoding processing device according to a third embodiment.

FIG. 9 is a diagram showing an example of combining reduced images in the embodiment shown in FIG.

10 is a diagram showing another example of division / combination of image information in the embodiment shown in FIG.

FIG. 11 is a diagram showing still another division / combination example of the image information in the embodiment shown in FIG. 1;

FIG. 12 is a basic configuration diagram of an image encoding / decoding processing device in a conventional example.

FIG. 13 is a diagram illustrating a problem of parallel processing in a conventional example.

[Explanation of symbols]

2 ... Scan order conversion unit, 4 ... Image information storage unit, 6 ... Image information division / synthesis unit, 8, 8 ', ..., 8 "... Image information coding / decoding unit, 10 ... Code information division / synthesis unit, 101 ... Image information dividing means, 102, 102 ', ..., 102 "... Encoding / decoding means, 103 ... Code information dividing / composing means

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification number Internal reference number for FI Technical indication H03M 7/40 8842-5J H04N 1/387 1/415

Claims (3)

[Claims] 1. An image information division / combination means for dividing / combining image information, a plurality of encoding / decoding means for encoding / decoding image information in parallel, and a code information division / composition means for dividing / combining code information. An image encoding / decoding processing method in the image encoding / decoding processing, wherein at the time of encoding, the input image information is divided by the image information division / combination means so that the statistical properties of the divided image information are close to each other. , Each of the divided image information is input to each of the plurality of encoding / decoding means to be encoded in parallel, and the plurality of code information output from each of the plurality of encoding / decoding means is combined in the code information division / combination means. To obtain combined code information, and at the time of decoding, the input combined code information is divided into a plurality of pieces of divided code information in the code information division combining means, and the plurality of pieces of divided code information are divided. Note that a plurality of pieces of encoding / decoding means are input to be decoded in parallel, and a plurality of pieces of decoded image information output from each of the plurality of encoding / decoding means are combined in the image information division / combination means to be combined image information. An image encoding / decoding processing method characterized by the following. 2. An n (where n is a natural number of 2 or more) encoding / decoding means is used, and only one element of n pieces of decoded image information output from the n encoding / decoding means is used. The image encoding / decoding processing method according to claim 1, wherein a reduced image of 1 / n of the original image is obtained by using the reduced image. 3. The image coding / decoding processing method according to claim 2, wherein n pages of the reduced image are combined to create an image.