JPH08316844A - Code data decoding method and code data decoder - Google Patents

Abstract

PURPOSE: To realize completely parallel decoding processing without an overhead with respect to code data of a variable length unable to be decoded from an optional position. CONSTITUTION: Serial decoding processing 12 by using a processor is conducted for code data in the first decoding of code data of a variable length, prescribed information specifying a segment position of the code data is generated in the unit of coding processing, and generated prescribed information is stored in a prescribed information storage file 15 in cross reference with the code data (processing 14). In the case of 2nd and succeeding decoding of the code data, prescribed information in cross reference with the code data is retrieved from the prescribed information storage file 15 and extracted (processing 11) and parallel decoding processing (13) by plural processors is conducted for each of partial code data divided by a section specified by the extracted prescribed information.

Description

Detailed Description of the Invention

[0001]

The present invention relates to a workstation,
Word processor, facsimile machine, printer machine,
The present invention relates to a technique suitable for decoding coded data such as a color still image that is compression-coded in a copying apparatus or the like.

[0002]

2. Description of the Related Art As an international standard compression encoding method for color still images, JPEG (Joint Photograhpic Coding Ex
perts Group) method. The JPEG method is roughly 2
There are two types.

The first method is DCT (Discrete Cosine).
Transform), which is irreversible in image restoration due to data quantization and the omission of high-frequency components of the image, but it is possible to realize a high compression rate while suppressing deterioration of image quality. It is the basic method of the JPEG method.

The second method is DPCM (Differen
tial Pulse Coded Modulation)
Although it has a low compression rate, it is a lossless method that can completely restore the original image. The DCT method is also an essential function Base
It is classified into line System and Extended System which is an optional function.

The Baseline System is a minimum necessary function that must be installed in all of the coding and decoding devices that employ the JPEG system, and the Huffman coding process is performed at the final stage of the coding process. Also, Extended Sy
The stem is an optional function prepared for a wider range of applications, and arithmetic coding processing is performed at the final stage of coding processing. In practice, Baseline
System can provide sufficient performance for most applications.

On the other hand, the functional specifications of the JPEG system (ISO DI
S 10918-1 REQUIREMENTS AND GUIDELINES)
The restart marker code (RSTm:
It is allowed to insert m = 0-7). JPEG
Basically, the coded data cannot be decoded unless the decoding process is performed serially from the beginning, but for the JPEG coded data in which the restart marker code is inserted, the restart marker code is detected and detected. The decoding process can be performed from the position of the restart marker code. This makes it possible to realize resynchronization and parallel decoding processing when a transmission error occurs in JPEG encoded data.

[0007]

However, the restart marker code is not specified to be always inserted. Therefore, for example, as such as a color printer equipped with a plurality of processors, for the decoding device do not know in advance what JPEG encoded data is input, the presence of restart marker codes can not be assumed.

[0008] For example, in the case of the Baseline System, the decoding process includes Huffman decoding and inverse DCT (Discrete Coding).
sine Transform) processing, inverse quantization processing, YUV → RGB
The processing is performed in the order of conversion processing. However, regarding the Huffman decoding processing, JP
Decoding from any position in the EG coded data is generally impossible.

Inverse DCT performed after the Huffman decoding process
Regarding the processing, the inverse quantization processing, and the YUV → RGB conversion processing, the position of the data to be processed in the image space and the number of data as processing units are clear, and the processing content is highly independent, so parallel processing is performed. Processing can be performed.

In other words, when performing JPEG decoding processing in parallel, the Huffman decoding processing that should be performed first becomes a bottleneck, and the point is how to perform this processing efficiently.

As a method of solving the above problems, there is the 49th National Convention of Information Processing Society of Japan: 7S-01: "Parallel processing method of JPEG compressed image decoding".

According to this method, in a decoding device having two processors, JPEG code data to be decoded is divided into first partial code data and second partial code data, and the first processor is Decoding is performed from the beginning of the first partial code data, that is, from the beginning of the code data, and the second processor starts the second partial code data, that is, in the middle of the code data (at an arbitrary position in the code data). ). Then, if there is a contradiction in the decoded data, the second processor repeats the trial-and-error process of restarting the decoding process from a new position, and the decoding process by the first processor becomes the second partial code data. It is a method of adopting the portion as the decoding result when the correct answer is included in the decoding result by the second processor when it arrives.

According to this method, the Huffman decoding process can be performed in parallel, but trial and error iterative processing of the decoding process by the second processor and confirmation of the validity of the decoding result by the second processor are performed. Since the overhead associated with etc. is unavoidable, it cannot be expected to perform completely parallel processing.

Further, even if the same JPEG code data is input again later, there is no choice but to perform trial and error processing as before, and no experience of decoding the same JPEG code data is utilized at all. There was a problem.

An object of the present invention is to perform completely parallel decoding without overhead for code data such as JPEG code data and other variable-length Huffman code data that cannot be decoded from any position in the code data. An object of the present invention is to provide a code data decoding method capable of realizing processing.

[0016]

In order to achieve the above object, a coded data decoding method provided by the present invention is provided with a plurality of processors and a storage device for the first time to obtain coded data of variable length to be decoded. At the time of decoding, the decoding process for the code data is serially performed by one processor, and at the same time, the predetermined information for specifying the break position of at least one code unit in the code data is generated and generated. The information is stored in the storage device in association with the ID of the code data, and when the variable-length code data to be decoded is decoded for the second time and thereafter, the information is stored in association with the ID of the code data. Extract the predetermined information stored in the device,
For each of a plurality of partial code data obtained by dividing the code data at a break position included in the extracted predetermined information, a decoding process for each of the partial code data is performed in parallel by a plurality of processors. .

Whether or not the code data is to be decoded for the first time is determined by determining whether or not the predetermined information associated with the variable-length code data to be decoded is stored in the storage device. Can be determined.

The break position of the code unit may be, for example, a break position of the Huffman code.

The ID of the code data may be the code data itself.

Further, the ID of the code data may be data which is a part of the code data or data which is a part of the code data. In this case, the ID of the coded data may further include a data attribute including at least one of the data name of the coded data, the data size, the creation date, and the image size after decoding.

[0021]

According to the present invention, for example, for JPEG code data in which the restart marker code is not inserted,
When decoding for the first time, the decoding process for the JPEG code data is serially performed by one processor, and at the same time, the predetermined information for specifying the break position of at least one code unit in the JPEG code data is generated. , The JP
It is stored in the storage device in association with the EG code data,
When the same JPEG coded data is decoded for the second time and thereafter, the predetermined information stored in the storage device in association with the JPEG coded data is extracted and the predetermined position is detected at the break position included in the extracted predetermined information. For each of a plurality of partial code data obtained by dividing the JPEG code data, the decoding process for each of the partial code data is performed in parallel by the plurality of processors, so that the JPEG code data that has been decoded in the past has been used. For no overhead,
It is possible to realize a completely parallel decoding process. Therefore, it is possible to effectively utilize the architecture of the decoding device having a plurality of processors and to speed up the decoding process.

[0022]

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

FIG. 1 is an explanatory diagram showing the basic concept of the decoding process of this embodiment.

In the figure, 10 is a determination process, 11 is a predetermined information search process, 12 is a first decoding process, 13 is a second decoding process, 14 is a predetermined information storing process, and 15 is a predetermined information storing file.

The decoding process of this embodiment is realized by a plurality of processors (not shown) executing a program.

The judgment processing 10 judges whether or not the predetermined information corresponding to the code data to be decoded exists in the predetermined information storage file 15. The predetermined information is the J to be decrypted.
The data is necessary for decoding the PEG coded data in parallel, and details thereof will be described later.

In FIG. 1, the solid line represents the flow of processing when it is determined that the predetermined information does not exist, and the dotted line represents the flow of processing when it is determined that the predetermined information exists.

First, the case where it is determined that the predetermined information does not exist (the flow of processing indicated by "solid line") will be described.

The first decoding process 12 decodes the code data to be decoded. The decoding process performed by the first decoding process 12 is a serial decoding process by one processor. Subsequently, the predetermined information storage processing 14 generates predetermined information from the code data to be decoded, associates the generated predetermined information with the code data to be decoded, and stores the predetermined information storage file 1
Store in 5.

Next, the case where it is determined that the predetermined information is present (the flow of processing indicated by "dotted line") will be described.

The predetermined information retrieval processing 11 searches the predetermined information storage file 15 for the predetermined information corresponding to the code data to be decoded.
To load from. Then, the second decoding process 13 decodes the code data to be decoded by using the loaded predetermined information. The decoding process by the second decoding process 13 is a parallel decoding process by a plurality of processors.

The coded data to be decoded may be, for example, coded original data such as a still image, a moving image, or audio, document image data output in a page description language, computer graphics, or the like. Various types are conceivable, such as image data and vector data used.

FIG. 2 is an explanatory diagram showing the basic concept of the second decoding process 13.

In the figure, 20 is a code division process, 21 is a first partial decoding process, 22 is a second partial decoding process, and 23 is an nth part.
Is a partial decoding process, and 24 is a decoding result merge process.

The code division processing 20 is a predetermined information retrieval processing 1
1 uses the predetermined information loaded from the predetermined information storage file 15 to divide the code data to be decoded into n pieces of partial code data, and divide the divided n pieces of partial code data into the first partial decoding data, respectively. Process 21, second partial decoding process 2
2, ..., Output to the nth partial decoding process 23.

Subsequently, the first partial decoding process 21, the second partial decoding process 22, ..., And the nth partial decoding process 23 respectively decode the partial code data output from the code division process 20.

Finally, the decoding result merge processing 24 is performed by the first
The partial decoding data 21, the second partial decoding process 22, ..., And the n-th partial decoding process 23 are merged into one decoded data.

As shown in FIGS. 1 and 2, in the decoding process of the present embodiment, the code data for which the predetermined information does not exist is serially decoded by one processor, and the code for which the predetermined information exists exists. The data is divided into a plurality of partial code data by using the predetermined information, and a parallel decoding process is performed by a plurality of processors for each of the divided partial code data.

Now, a decoding device to which the decoding process of this embodiment is applied will be described.

FIG. 3 is a block diagram showing the structure of a decoding device to which the decoding process of this embodiment is applied.

In the figure, 30 is a first processor, 31 is a second processor, 32 is an nth processor, and 33 is a first processor.
Local memory used by the processor 30 of the above, 34 is a local memory used by the second processor 31, 35 is a local memory used by the nth processor 32, 36
Is a first processor 30, a second processor 31, ...
It is a shared memory used by all of the nth processors 32.

Further, 37 is a JPEG code data storage file in which JPEG code data is stored, and 38 is a P in which predetermined information (hereinafter referred to as Profile data) is stored.
rofile This is a data storage file.

FIG. 4 is a block diagram showing the configuration of an information processing device to which the decoding device shown in FIG. 3 is applied.

In the figure, 40 is a processing unit comprising a plurality of processors, 41 is a main memory, 42 is an external storage device, 43 is a printer for printing out decoded data, and 44.
Is a display for displaying the decoded data, and 45 is a keyboard.

The information processing apparatus shown in FIG. 4 can be used as, for example, an image retrieval apparatus or a color printer apparatus.

Further, the first processor 3 in FIG.
0, second processor 31, ..., Nth processor 32
Corresponds to the processing device 40. In addition, the local memories 33, 34, 35 and the shared memory 36 in FIG.
Is a storage area secured on the main memory 42,
The JPEG code data storage file 37 and the Profile data storage file 38 in FIG.
It is a file recorded in 2.

Returning to FIG. 3, in the local memory 33,
Used by the first overall control area 33a, the first initialization area 33b, the first decoding area 33c, the Profile generation area 33d, the merge area 33e, and the first processor 30. It has a work area 33f.

A program for controlling the entire decoding process performed by the first processor 30 is stored in the first overall control area 33a, and the first initialization area 33 is stored.
In b, a program for setting data necessary for the first processor 30 to perform a decoding process, such as a decoding range of JPEG coded data to be decoded by the first processor 30, is stored. In the decryption area 33c, the first
A program for the processor 30 to perform the decoding processing is stored. Further, a program for converting the Profile data into a file is stored in the Profile generation unit area 33d, and a plurality of partially decoded data generated when parallel decoding processing is performed in the merge unit area 33e. Contains a program for merging.

The local memory 34 also includes a second overall control area 34a, a second initial setting area 34b, and
It has a second decoding unit area 34c and a work area 34d used by the second processor 31.

A program for controlling the entire decoding process performed by the second processor 31 is stored in the second overall control area 34a, and the second initial setting area 34 is stored.
b stores a program for setting the data necessary for the second processor 31 to perform the decoding process such as the decoding range of the JPEG coded data decoded by the second processor 31. In the decoding area 34c, the second
A program for the processor 31 to perform the decoding processing is stored.

The local memory 35 has an nth overall control area 35a, an nth initial setting area 35b, and
It has an nth decoding unit area 35c and a work area 35d used by the nth processor 32.

A program for controlling the entire decoding process performed by the nth processor 32 is stored in the nth overall control area 35a, and the nth initial setting area 35a.
In b, a program for setting data necessary for performing the decoding processing of the nth processor 32 such as the decoding range of the JPEG encoded data decoded by the nth processor 32 is stored. A program for the n-th processor 32 to perform a decoding process is stored in the sub-region 35c.

Further, the shared memory 36 has a shared work area 36a used by all of the first processor 30, the second processor 31, ...
Data required to generate e-data (hereinafter referred to as MCU
This is called data. ) Is stored in the MCU data output area 3
6b and. The details of the MCU data will be described later.

In this embodiment, as shown in FIG.
Although the JPEG code data storage file 37 and the Profile data storage file 38 are directly connected to the bus, for example, another information processing device connected to the information processing device to which the decoding device is applied via a network is a hard disk or the like. The recording medium may have the JPEG code data storage file 37 and the Profile data storage file 38.

Further, in the present embodiment, in order to simplify the explanation, the number of processors is n (n is a natural number), and the JPEG code data to be decoded is an MCU (Minimum) which is the minimum unit of code processing. It is assumed that Coded Unit) is not thinned out, but this does not lose generality, and is the same in principle when handling JPEG coded data that is thinned out by MCU.

FIG. 5 is a processing flowchart of the first processor 30.

As shown in FIG. 5, the first processor 30
Is a variable F secured in the shared work area 36a.
LG is initialized to "0" (step 500).

Subsequently, the JPEG code data designated by the user to be decoded is loaded from the JPEG code data storage file 37 into the shared work area 36a, and the Profile corresponding to the loaded JPEG code data is loaded.
It is determined whether the data exists in the Profile data storage file 38, and if it exists, the corresponding Profile data is loaded into the shared work area 36a (step 50).
1). In addition, the specification of JPEG code data and JP
The correspondence between the EG code data and the Profile data will be described later.

First, the case where there is no corresponding Profile data will be described.

When the corresponding Profile data does not exist (step 502), it is determined whether or not the restart marker code is inserted in the JPEG code data to be decoded (step 503). The presence / absence of the restart marker code depends on the JPEG functional specification (ISO DIS 10918-
According to the specifications described in 1 REQUIREMENTS AND GUIDELINES), it is possible to easily check the presence or absence of restart marker codes as well as their insertion intervals.

When the restart marker code is not inserted, the value of the variable FLG does not correspond to the JPEG code data to be decoded, and the corresponding Profile data does not exist, and
A value (for example, "1") indicating that the restart marker code is not inserted is set (step 504). Subsequently, serial decoding processing is performed by the first processor 30 (step 505). The details of step 505 will be described later.

When the restart marker code is inserted, the value of the variable FLG is set to the JPEG to be decoded.
The code data is set to a value (for example, "2") indicating that the corresponding marker data does not exist, but the restart marker code is inserted (step 506).
Then, the restart marker code is used to perform parallel decoding processing by the first processor 30, the second processor 31, ..., And the nth processor 32 (step 507). The details of step 507 will be described later.

In steps 505 and 507, since the MCU data is stored in the MCU data output area 36b as described later, Profile data is generated based on this MCU data (step 508). The details of step 508 will be described later.

Further, step 505 and step 50
7, the decrypted data is stored in the shared work area 3 as described later.
6a, the decoded data is stored in the printer 43.
Is printed out or displayed on the display 44 (step 509).

Finally, since the value of the variable FLG is not "3" (step 510), the Pr generated in step 508 is generated.
The ofile data is stored in the Profile data storage file 38 in association with the JPEG encoded data to be decoded (step 511).

Next, a case where corresponding Profile data exists will be described.

If the corresponding Profile data exists (step 502), the value of the variable FLG is set to the J to be decoded.
It is set to a value indicating that there is a Profile corresponding to the PEG coded data (for example, "3") (step 51).
2). Then, using the corresponding Profile data, the first processor 30, the second processor 31, ...
The parallel decoding process is performed by the processor 32 (step 513). The details of step 513 will be described later.

In step 513, as will be described later, the decrypted data is stored in the shared work area 36a, so the decrypted data is printed out by the printer 43 or displayed on the display 44 (step 509).

FIG. 6 is a detailed processing flowchart of step 505.

As shown in FIG. 6, the first processor 30
First, the decoding range of the JPEG coded data to be decoded is set (step 600).

Here, since the first processor 30 is a serial decoding process and only the first processor 30 performs the decoding process, the decoding range is set from the beginning to the end of the JPEG code data.

Subsequently, the first processor 30 decodes the JPEG coded data in the decoding range set in step 600, stores the decoded data as the decoding result in the work area 33f, and extracts the MCU data in the decoding process. Then, the extracted MCU data is stored in the MCU data output area 36b (step 601).

FIG. 7 is a detailed processing flowchart of step 507.

As shown in FIG. 7, the i-th (1.ltoreq.i.ltoreq.n) processor first sets the decoding range of the JPEG code data to be decoded (step 700).

Basically, the JPEG code data must be sequentially decoded from the beginning, but in order to realize resynchronization and parallel decoding processing when a transmission error occurs, up to 8 pieces of JPEG code data can be obtained. The specification allows the insertion of restart marker code. Therefore, by detecting the position of the restart marker code inserted in the JPEG encoded data and decoding each from the detected position, parallel decoding processing is performed for the JPEG encoded data in which the restart marker code is inserted. It can be easily realized.

For example, N (N ≦ 9) pieces of JPEG code data, such as the first partial code data, the second partial code data, ..., The Nth partial code data, depending on the inserted restart marker code. Consider the case of being divided into partial code data of.

When n = N, the decoding range of the first processor 30 is set from the beginning to the end of the first partial code data, and the decoding range of the second processor 31 is set to the second partial code. The data is set from the beginning to the end, and the decoding range of the nth processor 32 is set from the beginning to the end of the nth partial code data.

When n> N, the N + 1th processor,
Since the N + 2th processor ..., The nth processor 32 do not have to perform the decoding process, the N + 1th processor,
Nothing is set in the decoding range of the (N + 2) th processor ..., The nth processor 32.

When n <N, the nth processor 32
To decode the nth partial code data, the (n + 1) th partial code data, ..., And the Nth partial code data, the decoding range of the nth processor 32 is From the beginning to the end of the Nth partial code data (the end of the JPEG code data) is set.

Subsequently, the i-th processor proceeds to step 7
Decode the partial coded data in the decoding range set by 00, store the partially decoded data that is the decoding result in the corresponding work area, extract the MCU data in the decoding process, and output the extracted MCU data to the MCU data. Area 36
It is stored in b (step 701).

Subsequently, only the first processor 30 (step 702) merges the partially decoded data stored in the work areas 33f, 34d, 35d, respectively, and stores the decoded data as the merge result in the shared work area 36a. (Step 703).

FIG. 9 is an explanatory diagram showing the contents of MCU data.

As shown in FIG. 9, the MCU data is MC
Huffman code break position, which is the break of U, and MCU
Which composes Y (luminance) component, U (color difference) component, V
It is not the difference value of each DC data of the (color difference) component but the true value. The break position of the Huffman code is represented by the byte position and the bit position.

The break position of the Huffman code is used as a start position when the Huffman decoding process of the decoding process is performed in parallel. Further, the true value of the DC data is the data used when performing the processes after the Huffman decoding process in the decoding process.

In the JPEG code data, the Y component,
DC data of U and V components is the DC of the first MCU
Only the data is a true value, and the DC data of the second and subsequent MCUs is represented by the difference value of the immediately preceding value. Then, when the Huffman decoding process of the decoding process is performed, a process of sequentially calculating the true value of the DC data is also performed. Therefore, if the true value of the DC data on the way is obtained, it becomes possible to calculate the true value of the DC data subsequent to the DC data by using the position of the DC data as the start position, and the Hu of the decoding process is performed.
The processes after the ffman decoding process can be performed completely in parallel.

For example, the image size after decoding is 512 × 5.
Consider JPEG coded data with 12 pixels.

Since the JPEG coded data that is not thinned out by the MCU is coded in block units with 8 × 8 pixels as one block, the total number of blocks is 4096.
Since (= 64 × 64), the total number of MCU data entries is 4096.

In step 701, the first processor 30, the second processor 31, ..., The nth processor 32 decode the partial code data in parallel. The MCU data regarding the partial code data is stored in the MCU data output area 36b, the second processor 31 stores the MCU data regarding the second partial code data in the MCU data output area 36b, and the nth processor 32 is The MCU data relating to the partial code data of n is converted into the MCU data output area 36b
To be stored. Therefore, the MCU data will be stored in the MCU data output area 36b as n pieces of divided MCU data. For example, the first processor 30
Considering that the MCU data stored by the second processor 31 follows the MCU data stored by the above, it can be theoretically considered as one MCU data shown in FIG.

FIG. 10 is an explanatory diagram showing the contents of Profile data.

As shown in FIG. 10, the Profile data is
Each of the first processor 30, the second processor 31, ..., The nth processor 32 is composed of the start position of the partial code data to be decoded and the DC value of each of the Y component, the U component, and the V component. The start position is represented by the byte position and the bit position.

At step 513, as will be described later, the Profile data is used to convert the JPEG code data into
It can be decoded in parallel with no overhead.

FIG. 11 is a detailed process flowchart of the process of step 508.

As shown in FIG. 11, the first processor 3
For 0, the total number of blocks is calculated from the width W (dot) and the height H (dot) of the image size after decoding (step 1100).

Then, the number of blocks ΔB corresponding to the partial code data to be decoded by one processor is calculated (step 1101).

Subsequently, as shown in FIG. 12, from the MCU data stored in the MCU data storage area 36b, the 0th entry, the ΔBth entry, and the (2 × ΔB) th entry. , ..., ((n-1) × Δ
The B) th entry is decoded by the nth processor 32, the starting position of the partial code data to be decoded by the first processor 30, the starting position of the partial code data to be decoded by the second processor 31, and so on. Profile data is generated by extracting it as the start position of the partial code data to be generated, and the generated Profile data is stored in the Profile data storage file 38 (step 1102).

FIG. 8 is a detailed processing flowchart of step 513.

As shown in FIG. 8, the i-th (1.ltoreq.i.ltoreq.n) processor first sets the decoding range of the JPEG code data to be decoded (step 800).

Here, based on the Profile data shown in FIG. 10, as the decoding range of the i-th processor,
From the start position of the partial code data to be decoded by this processor to the start position of the partial code data to be decoded by the i + 1th processor. However, the nth processor 3
The second decoding range is set from the start position of the partial code data to be decoded by the nth processor 32 to the end of the JPEG code data.

Subsequently, the i-th processor proceeds to step 8
The partial coded data in the decoding range set by 00 is decoded, and the partial decoded data as the decoding result is stored in the corresponding work area (step 801).

Then, only the first processor 30 (step 802) merges the partially decoded data stored in each of the work areas 33f, 34d, and 35d, and the decoded data as the merge result is stored in the shared work area 36a. It is stored (step 803).

FIGS. 13 and 14 are explanatory views showing a method of associating the JPEG code data stored in the JPEG code data storage file 37 with the Profile data stored in the Profile data storage file 38.

The Profile data storage file 38 is centrally managed in a recording medium composed of one or a plurality of hard disks or optical disks.

The JPEG code data storage file 3
7 may be centrally managed like the Profile data storage file 38, but may be individually managed by a plurality of users in any unspecified recording medium such as a hard disk, an optical disk, a magnetic tape, a floppy disk, or a memory card. You may do it. That is, for example, a usage mode in which a plurality of users each store their own JPEG code data in an optical disc and carry it, and at the time of decoding, mount their own optical discs and load the JPEG code data to be decoded Can be considered.

When both the JPEG code data storage file 37 and the Profile data storage file 38 are centrally managed on the recording medium, the JPEG code data and Prof
It is easy to manage the ile data by associating it with the ile data.
You can manage the ofile data.

However, in the case where each of a plurality of users individually manages JPEG code data in an arbitrary unspecified recording medium, for example, the J with the same file name is given.
Since there may be multiple PEG coded data, associate it with the file name of the JPEG coded data, and
There is a problem that ile data cannot be managed.

FIG. 13 is an explanatory diagram showing an example of the association method for solving the above problem.

In FIG. 13, the JPEG code data stored in the JPEG code data storage file 37 is
Since it is managed individually by one user, there is no duplicate file name. Further, although the Profile data stored in the Profile data storage file 38 is centrally managed, although not shown, there may be duplicate file names.

Further, in FIG. 13, the solid line indicates the symbol input by the user to specify the JPEG coded data to be decoded (here, the file name "image A").
And the JPEG code data, and the dotted line represents the JPEG code data specified by the symbol input by the user and the corresponding Profile data.

In the example shown in FIG. 13, the Profile data is
The JPEG code data itself that generated the Profile data is managed as an ID. It should be noted that due to restrictions such as the capacity of the recording medium in which the Profile data storage file 38 is stored,
If the Profile data cannot be registered any more, the data that is considered unnecessary in the registered Profile data (for example, the oldest data) is deleted, and new data is registered instead. Management methods are also possible.

The user uses the keyboard 4 to specify desired JPEG code data from the JPEG code data stored in the JPEG code data storage file 37.
Input the symbol (“Image A”) from 5. The JPEG code data designated by the input symbol is transferred to the processing device 40 via the network or the bus. The processing device 40 recognizes the transferred JPEG code data (expressed as “1010001100” in the figure) as the ID of the Profile data, and searches the Profile data associated with the ID.

According to this method, the user simply specifies the desired JPEG code data without being aware of the presence or absence of the Profile data, and if the corresponding Profile data exists in the processing device 40, the parallel data is parallel. When the decryption process is performed and the corresponding Profile data does not exist, the serial decryption process is performed.

As an extension of the above method, instead of using the JPEG code data itself as an ID, it is also possible to use data that is a part of it or data that is a part of it as a plurality of IDs. In this case, by further combining with data attributes such as file name, file generation date, data size, and image size after decoding,
A highly reliable ID can be realized.

FIG. 14 shows Pr by the ID specified by the user.
It is explanatory drawing which shows the method of searching ofile data.

When generating the Profile data, the processing unit 40 automatically gives, for example, an unused minimum natural number as the ID of the Profile data, and displays the natural number given as the ID on the display 44. The user is notified of the ID of the Profile data corresponding to the decoded JPEG code data. If the Profile data cannot be registered any more due to restrictions such as the capacity of the recording medium in which the Profile data storage file 38 is stored,
A message indicating that registration is impossible is displayed on the display 44, while the user has previously sent his JPEG message.
ID assigned to Profile data corresponding to code data
Remove unwanted stuff from the
A data management method such as registration of ile data and acquisition of ID is also conceivable.

Therefore, the user can use the same JP after the second time.
When decoding the EG code data, a symbol (for example, “image A”) for designating the JPEG data is input from the keyboard 45, and the natural number (“2 )). The JPEG code data designated by the input symbol is transferred to the processing device 40 via the network or the bus. The input natural number is also transferred to the processing device 40 via the network or the bus. The processing device 40 uses the transferred natural number (“2”) as Pr.
It is recognized as the ID of the ofile data, and the Profile data associated with the ID is searched.

In the example shown in FIG. 14, Pr
The ofile data storage file 38 is managed separately for each user who uses it. This makes it possible to search the Profile data for each user by inputting a user name and password from the keyboard 45.

In the above-described embodiment, the partial decoded data is merged when the parallel decoding processing is performed. However, if the head block number of each partial decoded data is stored and held. , When the printer 43 prints out, the partial decoded data is written in the page memory at a position corresponding to the stored and retained block number, and when it is displayed on the display 44, it is stored in the frame memory. However, since it is possible to write the partially decoded data to a position corresponding to the block number stored and held, it is not necessary to merge.

[0118]

As described above, according to the present invention,
Even in the case of encoded data that cannot be decoded from the middle of coding, it is possible to perform completely parallel decoding processing without overhead, so that the architecture of a decoding device having multiple processors can be maximized. There is an effect that it can be effectively used and that the decoding process can be speeded up.

[Brief description of drawings]

FIG. 1 is an explanatory diagram showing a basic concept of a decoding method according to this embodiment.

FIG. 2 is an explanatory diagram showing a basic concept of second decoding processing.

FIG. 3 is a block diagram showing the configuration of a decoding device to which the decoding method of this embodiment is applied.

FIG. 4 is a block diagram showing the configuration of an information processing device to which the decoding device of this embodiment is applied.

FIG. 5 is a processing flowchart of a first processor.

FIG. 6 is a detailed processing flowchart of step 505.

FIG. 7 is a detailed processing flowchart of step 507.

FIG. 8 is a detailed processing flowchart of step 513.

FIG. 9 is an explanatory diagram showing the contents of MCU data.

FIG. 10 is an explanatory diagram showing the contents of Profile data.

FIG. 11 is a detailed processing flowchart of step 508.

FIG. 12 is an explanatory diagram showing a method for generating Profile data from MCU data.

FIG. 13 is an explanatory diagram showing an example of a method of associating JPEG encoded data with Profile data.

FIG. 14 is an explanatory diagram showing an example of a method of associating JPEG encoded data with Profile data.

[Explanation of symbols]

10 ... Predetermined information determination processing, 11 ... Predetermined information search processing, 1
2 ... 1st decoding process, 13 ... 2nd decoding process, 14 ... Predetermined information storage process, 15 ... Predetermined information storage file, 20 ...
Code division processing, 21 ... First partial decoding processing, 22 ... Second
Partial decoding process, 23 ... Nth partial decoding process, 24 ... Decoding result merge process, 30 ... First processor, 31 ... Second processor, 32 ... Nth processor, 33, 3
4, 35 ... Local memory, 36 ... Shared memory, 37 ...
... JPEG code data storage file, 38 ... Profile data storage file, 40 ... Processing device, 41 ... Main memory, 42 ... External storage device, 43 ... Printer, 44 ... Display, 45 ... Keyboard.

Claims (7)

[Claims] 1. When a plurality of processors and a storage device are provided and variable-length code data to be decoded is decoded for the first time, a decoding process for the code data is performed serially by one processor. , Generating predetermined information for specifying a break position of at least one code unit in the code data, storing the generated predetermined information in the storage device in association with the ID of the code data, and changing the variable to be a decoding target. When the long code data is decoded for the second time and thereafter, the predetermined information stored in the storage device in association with the ID of the code data is extracted, and the break position included in the extracted predetermined information is extracted. The code data is characterized in that, for each of a plurality of partial code data obtained by dividing the code data, a decoding process for each of the partial code data is performed in parallel by a plurality of processors. Decoding method. 2. The code data decoding method according to claim 1, wherein it is determined whether predetermined information associated with variable-length code data to be decoded is stored in the storage device. A coded data decoding method characterized by determining whether or not to decode the coded data for the first time. 3. The code data decoding method according to claim 1, wherein the break position of the code unit is a break position of a Huffman code. 4. The code data decoding method according to claim 1, 2 or 3, wherein the ID of the code data is the code data itself. 5. The code data decoding method according to claim 1, 2 or 3, wherein the ID of the code data is data which is a part of the code data, or a part of the code data is plural. A coded data decoding method characterized by being data. 6. The coded data decoding method according to claim 5, wherein the coded data ID is at least one of a data name, a data size, a creation date, and a decoded image size of the coded data. A coded data decoding method, characterized in that it includes a data attribute consisting of individual pieces. 7. Corresponding a plurality of processors each having a decoding means for decoding the coded data, the ID of the coded data and the predetermined information for specifying the break position of at least one coded unit in the coded data. A code data decoding device including a storage device for storing the data, wherein a specific one of the plurality of processors is associated with a predetermined length of code data to be decoded. Determination means for determining whether or not information is stored in the storage device, and decoding means for the one processor for performing decoding processing for the coded data when it is determined that the predetermined information is not stored. In the case where it is determined that the above predetermined information is not stored, the first control means for controlling serially
Storage means for generating predetermined information for identifying a break position of at least one code unit in the code data, and storing the generated predetermined information in the storage device in association with the ID of the code data; Is determined to be stored, the decoding process for each of the partial code data is performed for each of the plurality of partial code data obtained by dividing the code data at the break position included in the predetermined information. And a second control means for controlling the decoding means of the plurality of processors to perform the processing in parallel.