JPH04156182A - Picture processing system - Google Patents

Abstract

PURPOSE:To attain high speed extraction of a data corresponding to a part from an indefinite length compression data by providing an input section, a block position conversion section, a segmented position designation section, a segmented size storage section, a block address table BAT and a compressed data storage section CD to the system. CONSTITUTION:Since a block size to be segmented from a segmented size storage section 4 is inputted to a segmented position designation section 3, it is recognized that a range up to a block (m+2, n+1) from a block (m, n) as a head is to be segmented. Then at first a block address table BAT is accessed to obtain a head address (pointer) is obtained and it is read from a compressed data storage section CD and a head address in which a compressed data of a block (m+1, n) is described is obtained and its compressed data is read. Thus, the compressed data described address is obtained by the block address table BAT. Thus, the data is read at a high speed.

Description

[Detailed description of the invention] [Table of contents] Overview Industrial field of application Prior art (Figures 8 and 9) Means for solving the problem to be solved by the invention (Figure 1) Working examples (1) First example (Fig. 2) (2) Second example (Figs. 3, 4, and 5) (3) Third example (Fig. 6) (4) Fourth example Example (Figure 7) Effects of the invention [Summary] Regarding image processing method.

The purpose is to easily obtain the extraction position of compressed data using a block address e-table.

In an image processing method that performs encoding for each divided *@2o block and restores the compressed image, there is a compressed data holding means for holding compressed data for each two-stroke block, and a block indicating the entry position of the compressed data. An address e-table, a pull position conversion means that outputs a block position corresponding to the input specified position, and a block address table to obtain the entry position of the compressed data of the block to be cut out, and perform compression based on this. The present invention is characterized in that it includes a cutout position designation means for obtaining data.

[Industrial application field]

The present invention relates to an image processing method, and in particular to image compression and restoration in which encoding is performed for each divided block.

The present invention relates to an apparatus that enables compressed data for any part of an image to be extracted from compressed data at high speed.

[Conventional technology] In recent years, there has been a demand for creating a database of a single image.

Along with this, images can be efficiently compressed and stored.

There is a need for technology to quickly restore and display this accumulated information.

In general, image data has an extremely large amount of data compared to literary and numerical data, so efficient data compression technology is desired. Many coding techniques have been proposed, such as predictive coding, vector quantization, block coding, and transform coding, but transform coding is said to be particularly effective in terms of compression rate and quality. ing.

In transform encoding, one image is divided into an appropriate number of pixels (e.g. 8×
8) Divide into partial pictures * (blocks).

Each block is orthogonally transformed (fast Fourier transform).

After concentrating energy, the amount of data is compressed by quantizing the converted data. Therefore, the total amount of data is determined by the number of bits allocated to each spectrum of the converted data.

Efficient data compression is made possible by adaptively determining the number of bits to be allocated based on the statistical properties of the entire image.

There is a need for a technique for quickly restoring and displaying image data compressed using such a method.

As one of these technologies, a device has been developed that transfers compressed data as it is via a channel and quickly decompresses and displays it using the display's local function (for example, Japanese Patent Laid-Open No. 2-6
(See Publication No. 8673).

This device also has a local function that uses compressed data to perform processing such as enlarged display of two images and gradation conversion. However, the data compressed using the above method does not have a fixed length, but the size of each block is undefined. In order to extract the compressed data, the compressed data must be scanned from the beginning, and that process (: takes time.

Therefore, a technology is needed to quickly extract the target area without scanning the compressed data.

By the way, FIG. 8 shows adaptive discrete cosine transform coding, which is one type of transform coding.
Discrete Co51n Transform
This shows the format of image data compressed by coding:ADCT). First, the total number of bytes is written, and secondly, the size of each image in the X and Y directions is written. In this method, each block of an image divided into two is divided into five classes (C
o NCx;

That is, the variance value of each block is calculated and divided into 12 classes based on the histogram of the 2 variance values.

Bit allocation table B in compressed data shown in Figure 8
In T, the number of bits to be allocated to each element is written for each class. Note that this number of bits is fixed for each class, but the zero normalization table, which is different for each class, is compressed for each class by calculating the normalization coefficient based on the root mean square of the DCT transformed value for each class. It is something that exists.

The compressed image data CDo, CDi, etc. are written and arranged in blocks.

By the way, the compressed data of these images CDo, CD1...
・The size of the block belongs to the class (Co-C4
), and the size of the class is determined by the bit allocation table BT.

Therefore, when extracting a portion of one image, that is, a partial image, as shown in FIG. 9, it has been performed by scanning the compressed data in order from the first block, ie, Bo, B1, . . . .

The procedure will be explained next. First, as shown in FIG. 8, the bit allocation table BT is restored.

Each class Co = block data size BS of C4
Find (Cm) (n=Q, 1, 2.. 3, 4). Next, in FIG. 9 (2), the compressed data C of the first block Bo
The class number C is extracted from the flag written at the beginning of Do. In the example of FIG. 9(c), the class number is C.

Thus, the size BS (C6) of the compressed data of the block of class number Co obtained from the bit allocation table BT is obtained. And this size BS(Co
), the start address ADRo of the first block Bo
is added to the start address of the primary block Bl ADR
get l. Note that the size of the compressed data for each class is the sum of the number of bits for each element of the class in the bit allocation table ζ2.

That is, ADR1 = ADRo + B8 (Co) 0. Repeat this until the first block of the desired partial image,
When the first block is reached, a portion (Xo = XL) of the partial image is extracted in the same manner. Next, the compressed data is scanned again and the second row of data of one partial image is extracted.Such operations are repeated for the number of vertical blocks of the partial image to obtain compressed data of the partial image.

[Problem to be solved by the invention]

Therefore, for example, when a part of the image is extracted for processing such as enlarged dog display.

As mentioned above, the compressed data contains a mixture of multiple classes with different sizes.

There is a problem in that it takes time to scan these images, making it impossible to speed up the processing.The purpose of the present invention is to
An object of the present invention is to provide an image processing method that can quickly extract data corresponding to a portion from undefined length compressed data.

[Means for solving the problem] In order to achieve the above-mentioned object, the present invention provides two methods as shown in Fig. 1 (8).
As shown in the figure, the input section 1. Block position converter 2, cutout position designator 3. Cutting size holding section 4. It is equipped with a block address table BAT and a compressed data holding section CD.

In the present invention, FIG. 1 CB)! ': 1 display screen P as shown
is divided into multiple blocks. Figure 1 (5) shows an example of partitioning into N+1 blocks in the horizontal direction and N+1 blocks in the vertical direction.
It is configured with a size of x16 pixels.

In addition, the compressed data of each block is written in the compressed data holding unit CD, and the pointer indicating the start address of the compressed data writing light of each block is written in the block order in the block address table BAT. .

Note that the size of the partial image to be cut out is shown in block units in the cutout size holding unit 4;
When there are 3 blocks in the direction and 2 blocks in the X direction, the length of each block is entered.

[Effect]

Now, the operator uses, for example, a light pen.

When the point S shown in FIG. 1 is specified on the display screen P as the cutting destination position, the signal of this point S is input to the input section 1, and this is output as a coordinate signal in the X and X directions. The coordinate signal of this point S is input to the block position converter 2, and a block (m, n) of two points S is output.

Since the block size (3X2) to be cut out, that is, 3 blocks in the X direction and 2 blocks in the Y direction, is input to the cutout position specifying part 3 from the cutout size holding part 4, the blocks are (m+2.-n+1
) recognizes that the range up to This is read out from the compressed data holding unit CD and stored in the 9th block (m+1.
Find the start address where the compressed data 'n) is written and read out the compressed data.

In this way, a predetermined range of compressed data is sequentially read out and restored to obtain a partial image.

As described above, according to the present invention, since the compressed data writing light can be determined directly from the block address table BAT, these can be read out at high speed.

〔Example〕

(1) First Embodiment The first embodiment of the present invention will be explained based on FIG. 2. 0 In FIG. 2, the same reference numerals as in FIG. 2 is a block position conversion unit that receives a signal from the female mouth instruction part of a pen or mouse and outputs the coordinate signal; 2 is a block position conversion unit that converts the position according to the coordinate signal transmitted from the input unit 1; For example, when point S of block (m.

n). For this reason, two necessary data (to calculate the block position from coordinate signals such as block size (for example, 16x16), total image size, etc.) are held in this (Xru03 is the cutting position designation part, The block signal sent from the block position conversion section 2 and the cutout size holding section 4
The cutout size of the partial image transmitted from: Recognize the block position to be cut out based on the following.

This: Bi-based block address table BAT
is accessed to obtain the compressed data entry address of the desired block, and the compressed data of the desired block can be obtained from the compressed data holding unit CD.

The cropping size holding section 4 is used to record the size of a partial image to be cropped, when the partial image is to be cropped in the size of a block in the X direction and b block in the X direction.

It is written in the form a-b. For example, a=3. b=2 is entered.

5 is memory, block address table BA
The 0 block Φ address table BAT, which has a compressed data holding section CD, has the starting address of the memory in which the compressed data of each block is written.

As shown in FIG. 2(B), a tag TAG is added to the beginning of the block address table BAT, indicating that the block address table BAT is attached to the compressed data. In addition, the compressed data holding unit CD is
This is the same as the conventional compressed image data explained with reference to FIG.
(X direction).

(Y direction), a bit allocation table, a normalization table, compressed data for each block, etc.

A compressed data restoring section 6 restores the compressed data of the block read out from the compressed data holding section CD by the cutting position specifying section 3 and sends it to the display control section 7.

The display control section 7 performs control to display the restored data sent from the compressed data restoration section 6 on a display section 8 such as a CRT.

The display section 8 is for displaying, for example, an image.

Although it is composed of a CRT, a display such as a liquid crystal display may also be used.

The control unit 10 comprehensively controls the entire 2-image processing device, and is composed of, for example, a CPU, and performs control such as enlarging and displaying a partial image cut out from the compressed data storage unit CD. Note that the block address table BAT is as follows.

Generate it when compressing an image.

Next, the operation shown in FIG. 2 will be explained with reference to FIG. 1.

Now, when the operator indicates a point S on the display screen P shown in FIG. Output from 1.

Based on this, the block position conversion unit 2 converts the block (
m, n), and calculates this as the cutting position specifying unit 3.
Send to.

At this time, a=3. Since b=2 is entered, these numerical values are sent to the cutting position specifying section 3. Therefore, the cutout position designation section 3 has the above (m, n
) and a = 3, b = 2.
Blocks (m, n), (m+1*
”)y (m+2* “L <mp”+1
)+(m-1-1,n+1)t(m-4-2,r
It is recognized that t-4-1) is the range of the partial image to be read out.

First, the start address of the compressed data of block (m, n) is read from the block address table BAT, and based on this, the cutout position specifying unit 3 reads the compressed data of block (m, a) from the compressed data holding unit CD. The compressed data is read out, sent to the compressed data restoration section 6 to be restored to original data, and sent to the display control section 7 to display the restored image at a predetermined position on the display section 8. At this time, if the image is to be enlarged and displayed, the magnification is sent from the control unit 10, and the display control unit 7 performs enlargement processing based on this.

In this way, the block (m-)-1,n).

(m+2.n), (mt n+1), (m-4-
1, m+1) and (m+2.n+1), and one partial image can be cut out and displayed.

At this time, find the address of the first block of the first specified partial image using the block address table BAT,
This may be used as an offset from the beginning of the compressed data. A block corresponding to one horizontal line of the partial image is extracted from that address. Then, the address of the first block of the next row can be similarly obtained from the block address table BAT, and data for one block row can be taken out. By repeating this operation for the number of vertical blocks of the partial image, the desired partial image can be extracted at high speed.

As described above, in the present invention, since the start position of compressed data of each block can be obtained from the block e-address table, compressed data of a necessary partial image can be obtained at high speed.

(2) Second Embodiment The second embodiment of the present invention will be described based on FIGS. 3 to 5.
This will be explained with reference to FIGS. FIG. 3 is a schematic explanatory diagram of the second embodiment, FIG. 4 is a configuration diagram of the second embodiment, and FIG. 5 is an explanatory diagram of the operation of the second embodiment.

In the first embodiment, block address table B
Although the AT is stored in the database together with the compressed data, the block address table BAT may be generated from the conventionally read compressed data. This can be generated by scanning the entire compressed data once.

For example, as shown in FIG. 3(a), first, the partial image 11Dt
is extracted based on the conventional compressed data shown in FIG. 7, and at this time, a block address table is created for the scanned portion as shown in FIG. 3). When the next partial image is extracted as shown in FIG. 3(a), the block address table is used up to the part that has already been generated, as shown in FIG. 3(b). For other parts, the conventional compressed data is scanned and the block address table is updated.

FIG. 4 shows a configuration diagram of the second embodiment. In the figure, same as other figures.
The coded parts indicate the same parts.

Reference numeral 9 denotes a table creation/cutting processing unit, which creates a block address table BAT or reads compressed data, and handles each class Co=C.
A block size holding section 9-0 that holds the block size for each a, a display block instruction section 9-1 that instructs the block range to be displayed, and a block address table BA.
A BAT creation instruction-9 indicating the block position where T should be created is provided.

Next, the operation shown in FIG. 4 will be explained based on the flowchart shown in FIG. 5 and with reference to a map. An example in which the block size of 9 partial images is a:=4.1=4 will be explained.

■ Operator is not shown. Keyboard; enter commands. As a result, the control unit 1o.

Image compression data as shown in FIG. 8 is read from an external memory (not shown) and stored in the compressed data storage section CD of the memory 5.

(2) Next, the control section 10 reads the chiniru from the compressed data holding section CD and restores it in the table creation/cutting process B9. Then, the block size of the compressed data of each class Co=04 is determined from the allocation table 71, and this is entered in the block size holding section 9-0. In addition, the reblock size holding unit 9-0 has class Co (the size of one block of compressed data BS(o
) ~ Size of compressed data of block of class C4 BS (
4) is written in the block size holding section 9-0.

■ At this time, the block address table BAT has not yet been created and the first address of the compressed text for any block has not been entered, so EX=0. EY:=o is entered respectively. (EX, EY) is a block] ri・address・
The start address of the compressed data in table BAT indicates the last block written.

E ■ After inputting the command in (■) above, the operator must
Enter location. For example, specify the PI. This Plro
is transmitted to the input section l, and its coordinate signal is outputted to the block position conversion section 2;
Therefore, the first block x=Xl, y=Yo is output'
It can be done. As a result, the table creation/extraction processing unit 9.

a = 4゜〉 transmitted from the cutting size holding section 4
By b=4, the final block to be cut out (x=X4
.

y = ¥s), and calculate the first block (Xl,
, Yo) as well as the display block instruction section 9-
Fill in 1.

■ Then, the table creation/cutting processing unit 9 executes the
The display block instruction unit 9-1 checks the BAT creation instruction-9 to check whether the block address table B-AT for the first block (Xl, Yo) to be displayed has been created or not. .

This recognizes that BAT should be accessed from block (0,0), that is, that the block address table BAT for block (XIYo) has not been created yet.

■ As a result, the table creation/cutting processing section 9.

In order to create a block address table BAT from block (0,0), which is the first block written in BAT creation creation instruction-9, 7 lags are created from the compressed data of block (0,0) in the compressed data holding unit CD. co is read, and the compressed data length of the class Co, that is, the block size Be(o) is obtained from the block size holding unit 9-0.

Also block address fist table BAT block (
Write the start address ADRo of the compressed data of the book (0, 0) in the section 0°0). Enter BAT creation instruction-9 (secondary block (EX=1.BY=0)).

Next, by adding the block size BC(o) to the start address, the start address ADR1 of the compressed data of block (1,0) can be obtained.

Block this - address fist table BAT,
Fill in the classification of lock (1, 0), rewrite the BAT creation instruction-9, access the memory 5 using the start address, read 7 lags from the compressed data of block (1, 0), and create class C1. Then, the compressed data length of class C1 is obtained from the block size holding unit 9-0. By repeating this process, as shown in Figure 3), when blocks are configured, first the blocks (0, O), (
1, O) ~ (XE, O), (0, 1) ~ (XE,
1)...and.

Block Φ address table BAT is created sequentially,
In accordance with this, the BAT creation instruction-9 is sequentially rewritten.

■ When the block address table BAT for the block (Xs, Yo) is updated in this way and the BAT creation instruction-92 is rewritten, the table creation cutout processing section 9 uses the display block instruction section 9-1, this is the partial image Dl to be displayed.
Recognize that it is within the range of

■ As a result, the table creation/extraction processing section 9 converts the compressed data of block (Xl, Yo) into the compressed data decompression section 6.
This one block is restored and sent to the display control section 7 and displayed on the second display section 8. Partial image D1():) remaining 'J('
) 1 line (') 7 ' Tsuk (X2 * Yo)
e (Xs *Yo), (Xa, Yo), each block is sequentially restored and displayed, the block address table BAT is created, and the BAT creation instruction-9 is rewritten. Then, the undisplayed blocks (Xs, Yo) to (XE) are displayed again.

Block address table BAT for Yo )
is created and BAT creation instruction-9 is rewritten. In this way, the next row of blocks (0, Yt)
Block for the block of the undisplayed part about ~
0 and block (Xl, Y) where address table BAT is created and BAT creation instruction-9 is rewritten
t) Restoration and display of ~(Xa, Yt),
Block address - creation of table BAT and BAT
Creation creation instruction-9 is rewritten. When such processing has been performed up to the block (Xa, Ys), the table creation/cutting processing section 9 recognizes the completion of restoration of all blocks of the partial image D1 to be displayed by checking the display block instruction section 9-1. , the restoration process ends.

As a result, the block address table BAT has been created for the portion up to block (Xa, Ys). At this time, the BAT creation unit 9-2 has (Xs, Ys)
is entered.

In this state, the operator instructs P2 with a light pen in order to start losing the partial image @Dz. As in the above (■), the input section 1 converts the coordinate signal of Pl into the block position conversion section 24:
Output the first block X=Xo, y==
'(z is output. As a result, the table creation/cutting processing unit 9 receives a = a =
4, the final block ("" Xl #)' = Ys) to be cut out from b = 44 = is calculated, and these are written in the display block instruction section 9-1 along with the first block (Xo, Yz).

Next, in the same manner as in the above (■), the table creation/cutting processing section 9 performs the following steps.
In order to check whether the block address table BAT for the first block (Xo, Ya) to be displayed by the display block instruction section 9-1 has been created, the BAT creation instruction-9 is checked and 2 creation is performed. The block to do is (Xs.

Seeing that it is from the block (Xa, Ys),
Recognize that up to s) have already been created. Based on this, the first block of the two-part image D3 (Xo, Yz
) has been created.

Therefore, in the same manner as in the case (2) above, the table creation/extraction processing unit 9 accesses the block address table BAT to detect the leading position where the compressed data of the leading block (Xo, Ya) is written.

The compressed data is read from the compressed data holding section CD, and sent to the compressed data decompression section 6, where it is decompressed into aiijS, and sent to the display control section 7.

It is displayed on the display section 8. Such processing is partial image *D2
Block (Xo, Ya) ~ (Xs
.

Yz), and the second row block (Xo.

It is also performed for Ys) to (Ys, Ys).

However, since the pointer of the first block (Xa, Y4) on the third line has not been created in the block address table BAT, the table creation/extraction processing unit 9
In the same way as above, check the BAT creation instruction-9 and recognize that the block address should also create the table BAT from the block (Xs, Ys).

Ys ) ~ (X E, Ys ), create a block-address table BAT of blocks (o, Y4) ~ (X (1, Y4). Then, block (Xo,
Since Ya) should be displayed as 2, it is displayed on the display section 8 in the same manner as in the case of ■ and ■ above. In the same way, repeat T2y (Xl, Y4) ~ (Xs, Y4)
The block address table BAT is created while being displayed on the display section 8. and block (Xa,
For Y4) to (XE, Ya), a block address table BAT is created. This is also done for the last row of the partial image D2.

When the display and block address table creation processing for the social block (Xs, Ys) is performed, the table creation and cutout processing unit 9 recognizes that restoration of all blocks of the partial image has been completed, and the processing ends. At this time BAT
In the creation section 9-2, (X4, Ys) is entered as a block for which a secondary block address table is to be created.

As described above, according to the second embodiment of the present invention, while the first partial image is extracted using the conventional method, a block address table is created based on the scanned portion and the second partial image is extracted. When performing this, since the block address table for which the block address table has already been generated is used, it is possible to speed up the second and subsequent partial image retrieval processes.

(3) Third Embodiment In the first and second embodiments, addresses of all blocks are held in the block address table.

However, block address data requires a relatively large capacity, so if you need the size for all block addresses.

The compression ratio is not very large.

Therefore, in the third embodiment of the present invention, as shown in the sixth part, the block address table BATI is set only to the start address (ADH) of the block row, and
As shown in c), block address table BAT
2 may be used only for the start address (ADH) of the block row and the middle address (ADRyV/2) of the row. In FIG. 6(A), the black dots (BJ) indicate the positions written in the block address table.

Furthermore, as for the expression of block addresses, in addition to expressing all blocks as offsets from the beginning of the compressed data as described above, as shown in FIG. 6 (q).

Only the start address of the block line indicated by a black dot may be used as an offset from the start of the compressed data, and other blocks indicated by an x may be represented by the offset (OFF) from the start address (ADH) of that block line.

(4) Fourth example A fourth embodiment of the present invention will be described with reference to FIG.

Now, when a two-part image DI is displayed on the display unit as shown in the seventh position, for example, in a medical image such as an X-ray drawing, a range M as shown in the seventh position is displayed.

The partial image is often scrolled to 1/2 the top, bottom, left and right of the partial image @Dx. Here scroll means 1 part i
The purpose is not to move jj@Ih itself, but to display another partial image of the same size as the two-part image Ds.

Therefore, when restoring the two-part image D1, as explained in the second embodiment, the conventional format compressed image data shown in FIG. 8 is scanned to create a block access table in the scanning area S. However, the range required for the second step is range M, so in the seventh scan, among the scanning area S indicated by the dashed line, only the block address of the area BAT indicated by the right diagonal line is left, and the others are discarded. create a block address table. As a result, the address of the compressed data at the left end of the partial image displayed nine times is:

This partial block-address is obtained.

As the second partial image D2, as shown in FIG. 7(B),
When displaying a part of something that moved one block diagonally downward to the left, the address of the left end of the part excluding area S2 is 2 part 68
This can be easily obtained from the block address table of the area BAT portion created when scanning Dt.

Therefore, in order to display the second partial image D2, the block address table may be updated by scanning the areas 81 and 82.

In this way, the size of the block address table can be further reduced.

〔Effect of the invention〕

According to the present invention, the position of the desired compressed data can be easily obtained using the block address table, so scanning of the compressed data can be omitted or simplified and a part of the image can be extracted, so that two-part display and enlarged display can be performed. Image processing such as can be performed at high speed.

[Brief explanation of drawings]

FIG. 1 is a diagram showing the principle configuration of the present invention. FIG. 2 is an explanatory diagram of the first embodiment of the present invention. FIG. 3 is a schematic explanatory diagram of a second embodiment of the present invention. FIG. 4 is a configuration diagram of a second embodiment of the present invention. FIG. 5 is an explanatory diagram of the operation of the second embodiment of the present invention. FIG. 6 is an explanatory diagram of a third embodiment of the present invention. FIG. 7 is an explanatory diagram of a fourth embodiment of the present invention. FIG. 8 is an explanatory diagram of the format of image compression data. FIG. 9 is an explanation of the conventional method. 1...Input section. 2...Block position conversion section. 3... Cutting position designation section. 4... Cutting size holding section.

Claims (2)

[Claims] (1) In an image processing method that compresses and restores an image by encoding each divided image block, a compressed data holding means (CD) holds compressed data for each image block.
a block address table (BAT) that indicates the input position of the compressed data; a block position conversion means (2) that outputs the block position corresponding to the input instruction position; 1. An image processing method comprising: a cutout position specifying means (3) for obtaining the entry position of the compressed data of the block to be cut out, and obtaining the compressed data based on this. 2. The image processing method according to claim 1, further comprising: (2) table creation/cutting processing means (9), wherein said block address table (BAT) is generated from compressed data.