JPH05260311A - Method for expanding compressed image data - Google Patents

Abstract

PURPOSE:To obtain an image having the desirable picture quality without expanding the compressed image data over an entire screen for hierarchy when the compressed image data coded progressibly are expanded. CONSTITUTION:The original image data are subjected to the DCT conversion and the conversion coefficient is divided into quadripartite bands B1-B4. Then these bands B1-B4 are turned into the Huffman codes. Thus the compressed image data are obtained. When these compressed data are expanded, the compressed image data are first expanded on the entire screen for the first band B1. When the areas are designated for the extraction of the compressed image data in regard of the second and subsequent bands. Thus the compressed image data of the designated areas only are extracted and expanded. Then the picture quality is improved in the image data extracting areas.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for decompressing compressed image data, and more particularly to a method for extracting and decompressing a part of compressed image data generated by progressive coding.

[0002]

2. Description of the Related Art Image data for printing plate making has a feature that the data amount thereof is much larger than the image data of a still image displayed on a monitor. For example, the data amount of a still image displayed on the monitor is about 1 MB per screen, while the image data for printing plate making reaches about 60 MB per A4 sheet (the resolution is 400 lines / inch). If). To store such enormous amount of image data as it is, an enormous memory capacity is required, and it takes a lot of time to transfer the data. Therefore, an image data compression method is generally used in which the amount of data is reduced by encoding the image data and compressing the image data.

Techniques such as so-called vector quantization coding and orthogonal transform coding are used as a method of compressing multi-valued image data. Discrete cosine transform (hereinafter referred to as "DCT transform") and Hadamard transform are known as the orthogonal transform. At the time of compression, the transform coefficient obtained by the series transform such as DCT transform is quantized, and the transform coefficient is further encoded by entropy coding such as Huffman coding. These compression methods are capable of compressing image data at a high compression rate, but on the other hand, the image data obtained by restoring compressed image data does not completely match the image data before compression, so-called It is a lossy encoding method.

By the way, as a method of compressing image data suitable for image data retrieval and communication, there is known a progressive coding method for hierarchically compressing image data. An example of the progressive coding is the DCT progressive coding adopted in the international still image coding standard (JPEG).

FIG. 1 is an explanatory diagram for explaining an encoding method called the ss (spectral-selection) method in the JPEG method progressive encoding. When compressing an image by this ss method, first, referring to FIG.
The image data is DCT-transformed for each pixel block of 8 × 8 pixels in (a) to obtain 64 transform coefficients, and the 64 transform coefficients are converted into four bands B1 to B4 as shown in FIG. 1 (b). To divide. Each of these bands B1 to B4 corresponds to one layer. The frequencies of the alternating-current components included in the first to fourth bands B1 to B4 are gradually increasing, and therefore, finer image components are included in the order of the bands B1 to B4. When displaying the image on the monitor, first, as shown in FIG. 1C, the data of the first band B1 is expanded to display a rough image over the entire screen,
Next, when the data of the second to fourth bands are expanded as shown in FIGS. 1D to 1F, the image quality on the monitor is gradually improved.

Generally, image data compressed by progressive coding is composed of a plurality of layers of compressed image data whose resolution and gradation level are successively improved. Then, when the compressed image data of each layer is sequentially decompressed and the image is displayed on the monitor, an image with low resolution and gradation is first displayed over the entire screen,
The resolution and gradation are gradually improved. Then, the compressed image data of the entire screen is expanded for each layer until the displayed image has a desired image quality.

[0007]

As described above, the prior art is as follows.
When decompressing the progressively encoded compressed image data, the compressed image data of the entire screen is decompressed for each layer, so that there is a problem that it takes a considerable time to obtain a desired image quality. The present invention has been made to solve the above-mentioned problems in the prior art, and it is possible to obtain compressed image data of a desired image quality without decompressing the compressed image data of the entire screen for each layer. An object is to provide a decompression method.

[0008]

In order to solve the above-mentioned problems, a method according to the invention described in claim 1 is a method for decompressing compressed image data composed of a plurality of hierarchical data. a) displaying the entire image by decompressing the compressed image data of the first layer, and (b)
Specifying the next layer to be the target of the extraction process and designating the extraction region from which the compressed image data of the layer to be processed is to be extracted in the displayed whole image, and (c) the process target A step of extracting compressed image data corresponding to the extraction area from the compressed image data of the layer; and (d) extracting by decompressing the compressed image data of the layer to be processed extracted in the step (c). Updating the image of the region and displaying the entire updated image. In the method according to claim 2, the image data is serially converted for each pixel block in the entire image along a scanning line in a predetermined direction, and the conversion coefficient data obtained by the serial conversion is divided into a plurality of layers. A method of dividing and compressing a part of compressed image data from compressed image data created by entropy-encoding the transform coefficient data for each layer, the method comprising: A step of displaying the entire image based on the compressed image data of the first layer, and (b) specifying the next layer to be the target of the extraction processing, and also identifying the next layer to be processed in the displayed entire image. A step of designating an extraction area from which the compressed image data is to be extracted, and (c) sequentially decoding and decoding the entropy coded data of the compressed image data of the layer to be processed. Accumulating the number of the obtained data, and extracting the entropy coded data corresponding to the pixel block included in the extraction region based on the number of the decoded data, and (d) the step (c). Updating the image of the extraction area by decompressing the entropy-encoded data extracted in step 1, and displaying the entire updated image.

[0009]

According to the method of claim 1, after the whole image is displayed by decompressing the compressed image data of the first layer, the extraction region is designated, and the extraction region is selected from the compressed image data of the layer to be processed. Since the compressed image data corresponding to is extracted, decompressed, and the image is updated, the image quality of the extraction region can be improved. Therefore, it is possible to display an image of a desired image quality without expanding the compressed image data of the entire screen for each layer.

In the method of the second aspect, the data obtained by decoding the entropy coded data corresponds to the transform coefficient obtained by serially transforming the image data. Since the number of transform coefficients for each pixel block in each layer is predetermined, it is possible to extract entropy-coded data corresponding to the pixel blocks included in the extraction region based on the number of decoded data. Further, since the entropy-coded data (that is, the compressed image data) of the extracted area thus extracted is expanded to update the image, the image quality of the extracted area can be improved. Therefore, it is possible to display an image of a desired image quality for each layer without decompressing the compressed image data of the entire screen.

[0011]

EXAMPLES A. Overall Configuration of Apparatus FIG. 2 is a block diagram showing an image processing system as an embodiment of the present invention. The image processing system includes an image search device 10, an image input device 38, and an image data supply device 40. The image input device 38 is, for example, a reading scanner that reads image data of a document.

The image retrieving apparatus 10 comprises a CPU 12 and a bus line 14, and the bus line 14 has a memory 16, a decompression processing section 18, a smoothing processing section 20, a magnetic disk 22 and an input. An interface 24, a communication interface 30, and a frame memory 32 are provided. A mouse 26 and a keyboard 28 are connected to the input interface 24. Also, the frame memory 3
2, a D / A converter 34 and a CRT 36 are sequentially connected.

The image data supply device 40 includes a CPU 42.
And a bus line 44. The bus line 44 includes a memory 46, a compression processing unit 48, a compressed image extraction unit 50, an interface 58, and a communication interface 6
0, the magneto-optical disk 62, and the magnetic disk 64 are connected. The image data read by the image input device 38 is compressed by the compression processing unit 48, and the obtained compressed image data is stored in the magneto-optical disk 62 or the magnetic disk 64.

Image retrieval device 10 and image data supply device 4
0 are mutually connected via the respective communication interfaces 30 and 60. The operator is the image search device 1
When data specifying an image to be searched (hereinafter, referred to as “search input data”) is input to 0, the search input data is transmitted from the image search device 10 to the image data supply device 40. Compressed image extraction unit 50 of image data supply device 40
Reads the compressed image data stored in the magneto-optical disk 62 or the magnetic disk 64 according to the search input data, and transfers the compressed image data to the image search device 10. The decompression processing unit 18 of the image search device 10 decompresses the sent compressed image data to create bitmap data, and writes the bitmap data in the frame memory 32. As a result, the searched image is displayed on the CRT 36.

B. Structure of Compressed Image Data Before explaining the method of extracting the compressed image data, an outline of the method of compressing the image data and the structure of the compressed image data will be described. In this embodiment, s- of the international standard method for color still image coding (hereinafter referred to as "JPEG method") is used.
The image data is compressed according to the progressive coding method called the s method. Regarding this international standard system, "Outline of International Standard System for Color Still Image Coding (JPEG) (Part 1)", Takao Omachi et al., IEICE, 19
1991, Vol. 20, No. 1, pp. 50-58. For the s-s method, see page 53, right column to fifth page.
See page 4, left column.

FIG. 3 is a block diagram showing the structure of an encoding unit included in the compression processing unit 48. The coding unit has a DCT transform unit 72, a quantization unit 74, and a Huffman coding unit 76. The Huffman coding unit 76 also includes a DC coefficient coding unit 78, an AC coefficient coding unit 80, and a compressed data editing unit 82. DC
In the T conversion section 72, the original image data Do read by the image input device 38 is two-dimensionally DCT converted. At this time, the entire image region is divided into pixel blocks of 8 × 8 pixels, and DCT conversion is performed for each pixel block. In the case of a color image, each color plate (for example, each color plate of yellow, magenta, cyan, and black) is converted.
FIG. 4A shows a pixel block P in which the entire image IP is M × N.
FIG. 4 is a conceptual diagram showing a state of being divided into B, and FIG. 4B shows one pixel block PB in an enlarged manner.

Image data of 8 × 8 pixels (= 64 pixels) in one pixel block PB is converted to 8 by DCT conversion.
It is converted into × 8 conversion coefficients. FIG. 5 is an explanatory diagram showing the conversion coefficient F (p, q) obtained by the DCT conversion. Of the 8 × 8 transform coefficients F (p, q), F
(0,0) is a direct current component (hereinafter referred to as "DC coefficient"), and other than F (0,0) is an alternating current component (hereinafter referred to as "AC coefficient").

The quantizing unit 74 has 64 transform coefficients F.
Linearly quantize (p, q). FIG. 6 is an explanatory diagram showing an example of a quantization table used at the time of quantization. Each transform coefficient F (p, q) is assigned to each position (p, p,
Linear quantization is performed according to the following equation with the quantization width Ss (step size) shown in q). f (p, q) = round [F (p, q) / Ss (p, q)] (1) Here, f (p, q) is a quantized conversion coefficient, and the operator “round” "Indicates an operation for converting to the nearest integer.

The quantized transform coefficient f (p, q) is coded by the Huffman coding unit 76 and output as compressed image data Dc. In the JPEG method, the DC coefficient and the AC coefficient are encoded separately. FIG. 7 is a block diagram showing the function of the DC coefficient coding unit 78. Further, FIG. 8 is an explanatory diagram showing the contents of data processing in the DC coefficient encoding unit 78. Block delay unit 102 and adder 1
04 is the difference between the DC coefficient Fd of each pixel block and the DC coefficient Fd of the immediately preceding pixel block (hereinafter referred to as “DC difference”).
ΔFd is calculated. This is shown in Figures 8 (a) and (b).

The grouping processing unit 106 obtains the group number SSSS and the additional bit data Ad corresponding to the DC difference ΔFd according to the grouping table shown in FIG. 9 (see FIG. 8F). In FIG. 8, the additional bit data Ad is shown in binary. Group number SS
As shown in FIG. 9, SS is a number determined corresponding to the range of the DC difference ΔFd, and the additional bit data Ad
Is the DC difference specified by the same group number SSSS.
It is data indicating the smallest value in the range of Fd.

The group number SSSS is further Huffman-encoded by the one-dimensional Huffman encoder 108 to generate a group number code Hd (see FIG. 8 (d)).
FIG. 10 is an explanatory diagram showing an example of the Huffman code table 110 used by the one-dimensional Huffman coding unit 108. The Huffman code table 110 of FIG.
It is a table showing the relationship between the group number SSSS of (c) and code data Hd. The code data (codeword) obtained by the Huffman coding has a variable length as can be seen from FIG.

The code data Hd and the additional bit data Ad thus created are DC coefficient coded data CFd (see FIG. 8).
(See (g)) to the compressed data editing unit 82 (FIG. 3). As described above, the DC coefficient is converted into the code data CFd obtained by Huffman coding the DC coefficient difference ΔFd between the adjacent pixel blocks.

FIG. 11 is a block diagram showing the function of the AC coefficient coding unit 80. The AC coefficient Fa is first rearranged in one dimension by the zigzag scanning unit 112. Figure 1
2 is an explanatory diagram showing a route of zigzag scanning. Further, FIG. 13A shows an example of DCT coefficients arranged in 8 rows and 8 columns, and FIG. 13B shows AC coefficients rearranged by the zigzag scan. As shown in FIG. 13A, in this embodiment, the band B having four DCT coefficients is used.
It is divided into 1 to B4.

The determination unit 114 is arranged in the one-dimensional rearranged A
It is determined whether or not the value of the C coefficient Fa is 0. If the AC coefficient Fa is 0, the run length counter 116 converts consecutive AC coefficients of 0 into a run length count NNNN. If the AC coefficient Fa is not 0, the grouping unit 118
Thus, the group number SSSS and the additional bit data Aa are generated in the same manner as the DC coefficient. In FIG. 13 (c),
Eight non-zero AC coefficients have been converted to group numbers.

The run length count NNNN and the group number SSSS are Huffman-encoded by the two-dimensional Huffman encoder 120 to obtain code data Ha (see FIG. 13).
(D)) is generated. However, the Huffman code of the run length count NNNN is omitted in FIG. FIG. 14 is an explanatory diagram showing an example of the Huffman code table 122 used by the two-dimensional Huffman coding unit 120. This Huffman code table 122
Is a table showing the relationship between the data arranged as shown in FIG. 13C and the code data Ha.

The code data Ha and the additional bit data Aa thus created are given to the compressed data editing unit 82 (FIG. 3) as AC coefficient coded data CFa. The compressed data editing unit 82 edits the DC coefficient coded data CFd and the AC coefficient coded data CFa to create the compressed image data Dc of the image IP. At this time, the encoded data is collected for each of the bands B1 to B4 shown in FIG. 13A, and compressed image data for each band is generated.

FIG. 15 is an explanatory diagram showing the structure of the compressed image data Dc. Compressed image data Dc (i) for each band
The header part of the file name and the image size Wx, Wy
, Band division data, band number, and next band address are included. The image sizes Wx and Wy are the number of pixels of the image IP in the sub scanning direction and the main scanning direction, respectively. The band division data is the DCT included in the band.
It is data indicating the position coordinates of the conversion coefficient. The band number is a number indicating which of the four bands. The next band address indicates the address where the compressed image data of the next band is stored.

As shown in FIG. 15, the compressed image data of each band has the following features. (1) Compressed image data Dc (1) of the first band B1
Is the DC coefficient coded data CFd for each pixel block.
And AC coefficient coded data CFa are collected into one set. (2) In the bands after the second band B2, only the AC coefficient coded data CFa is grouped into one set for each pixel block. (3) Between DC coefficient coded data CFd and AC coefficient coded data CFa, and AC coefficient coded data CF
Data for identifying a data boundary is not inserted between a and encoded data is arranged continuously. (4) No data for identifying the boundary of the pixel block is inserted between the set of encoded data for each pixel block. (5) Since the variable length coding is performed, the data length of the coded data of each pixel block is different.

When the image data includes a plurality of color components (eg, four YMCK components), there are two methods of arranging the color components in the compressed image data Dc (i) of each band: non-interleave and interleave. is there. In non-interleaving, coded data is arranged for one color component as shown in FIG. 15, and then coded data for the next color component is continued. On the other hand, in interleaving, the coded data of each component of YMCK is arranged for each pixel block, and then the coded data of the next pixel block is continued. In this embodiment, a case will be described in which compressed image data generated according to non-interleave is extracted.

C. Overall Procedure of Search Process FIG. 16A is a diagram showing an image IP to be searched in this embodiment, and FIG. 16B shows a state in which the image IP is divided by pixel blocks PB. .. As described above, the image data of the image IP is first DCT-transformed for each pixel block, the transform coefficient is Huffman-coded, and the compressed image data Dc (i) as shown in FIG. 15 is divided into bands. Is created and stored in the magnetic disk 64 of the image data supply device 40.

FIG. 17 is a flow chart showing the procedure of a search process for compressed image data. In step S1, the parameter i representing the band is set to 1. Step S
In 2, the operator inputs data (search input data) for designating an image to be searched by the image search device 10. In step S3, the search input data is supplied from the image search device 10 to the image data supply device 40, and the compressed image data Dc (1) of the first band of the specified image is read from the magnetic disk 64 and the image is read. It is transferred to the search device 10. In step S4, the parameter i is incremented by 1.

In step S5, the compressed image data Dc (i) of the first band is decompressed by the decompression processing unit 18, the decompressed image data is thinned, and the reduced image is displayed on the CRT 36. .. Since the image displayed at this time is reproduced only by the compressed image data Dc (1) of the first band, the image has a considerably low image quality.

In step S6, the operator operates the CRT 3
The image displayed in 6 is observed, and it is determined whether or not there is a region where the image quality needs to be improved. Whether or not the image quality needs to be improved depends on the purpose of the search. For example, when a certain image is displayed on the CRT 36 and it is confirmed that the image is a desired image, when all the compressed image data of the image is transferred from the image data supply device 40 to the image search device 10, the image is displayed on the CRT 36. It suffices if it is possible to determine whether or not the selected image is the desired one, so the search ends when the image is displayed with a relatively low image quality. On the other hand, CRT3
When a part of the image displayed in 6 is cut and pasted on another image, a considerably high image quality is required for a desired area in the image.

When the process for improving the image quality of the entire image is designated in step S7, the compressed image data Dc of the i-th band for the entire image is specified in step S8.
(I) is the image data supply device 40 to the image search device 10
Transferred to. Then, the process returns to step S5 via step S4, the transferred compressed image data is decompressed, and the image is displayed again on the CRT 36. Since the image data component based on the compressed image data Dc (i) of the i-th band is added to the redisplayed image, the image quality of the entire image is improved.

When the process for improving the image quality of a partial region in the image is selected in step S7, the operator further determines the region where the image quality should be improved in step S9 (hereinafter referred to as "image quality improving region"). Is specified. For example, as shown in FIG. 16A, the first image quality improvement region R1 is designated by inputting the positions of the points P1 and P2 with the mouse 26, and the second image is input by inputting the positions of the points P3 and P4. The image quality improvement area R2 is designated.

The i-th compressed image data Dc (i) is extracted in the extraction regions ER1 and ER2 shown by hatching in FIG. 16 (b). FIG. 18 is an explanatory diagram showing the relationship between the image quality improvement region R1 and the extraction region ER1. In FIG. 18, (x, y) are pixel coordinates, and (m,
n) is a pixel block coordinate. Here, y and n are coordinate axes in the main scanning direction, and x and m are coordinate axes in the sub scanning direction.
Hereinafter, a case where the extraction region ER1 is mainly processed will be described.

In FIG. 18, the operator has two points P
When the positions of 1 and P2 are designated by the mouse 26, the coordinates (x1, y1), (x2, y2) of these points are given to the CPU 12. The CPU 12 uses these coordinates (x1, y
1), based on (x2, y2), the coordinates (m1, n1) of the pixel blocks PB1, PB2 including the points P1, P2,
(M2, n2) is calculated according to the following formula. m1 = INT (x1 / Px) +1 (2) n1 = INT (y1 / Py) +1 (3) m2 = INT (x2 / Px) +1 (4) n2 = INT (y2 / Py) +1 ((4)) 5) Here, the function “INT” indicates an operation of truncating the value in parentheses after the decimal point. Further, Px and Py indicate the numbers of pixels in the x direction and the y direction of the pixel block, respectively.

Image quality improvement region R1 designated by the operator
18 is a rectangular area having two points P1 and P2 as vertices on a diagonal line, as shown in FIG. On the other hand, the extraction area ER1 from which the compressed image data Dc (i) is extracted is an area including all pixel blocks including at least the image quality improvement area R1. In FIG. 18, the extraction area E
R1 is surrounded by a thick solid line.

Returning to FIG. 17, in step S10, the compressed image data Dc (i) of the i-th band for the extraction regions ER1 and ER2 is extracted by the compressed image extraction unit 50 in the image data supply device 40, and the image retrieval device Transferred to 10. Details of the extraction processing of the compressed image data in step S10 will be described later.

For each of the extraction regions ER1 and ER2, the compressed image data Dc (i) of the i-th band is the image retrieval device 1
When it is transferred to 0, it goes through step S4 to step S4.
5, the extracted compressed image data Dc (i) is expanded and the reduced image on the CRT 36 is updated. As described above, first, by repeating steps S4 to S8, the entire image is raised to a certain level of image quality, and after that, steps S9, S10, and S4 to S6 are repeated, so that the image quality of each part in the image is improved. Can each be raised to the desired level. The smoothing processing unit 20 may smooth the image after the extraction of the compressed image data Dc (i) is completed. The smoothing is performed to eliminate image distortion (called “block distortion”) between pixel blocks that occurs when the DCT coefficient is decoded.

D. Configuration and Operation of Compressed Image Extraction Unit FIG. 19 is a block diagram showing the internal configuration of the compressed image extraction unit 50 that extracts compressed image data. The compressed image extraction unit 50 includes a first control unit 202 and a separation circuit 204.
It has an AC decoding circuit 206, a DC decoding circuit 208, a counter 210, a second control unit 212, a bit counter 214, and a buffer 216. The separation circuit 204 has a function of separating the DC coefficient coded data CFd and the AC coefficient coded data CFa included in the compressed image data and transferring them to the DC decoding circuit 208 and the AC decoding circuit 206, respectively. However, in this embodiment, it is the compressed image data of the second and subsequent bands that are extracted for the image quality improvement region, and the compressed image data of the second and subsequent bands are DC coefficients as shown in FIG. Does not include encoded data. Therefore, in the extraction processing of the compressed image data in this embodiment, the compressed image data given to the separation circuit 204 is always transferred to the AC decoding circuit 206.

FIG. 20 is a flow chart showing the procedure of the extraction process executed by the compressed image extraction unit 50 in step S10 of FIG.

In step S30, the extraction area offset data transferred from the image retrieval apparatus 10 side and the band number indicating the band to be extracted are given to the first control section 202. Here, the extraction region offset data is data indicating the maximum value and the minimum value of the pixel block coordinates in the main scanning direction and the sub scanning direction of the extraction region, and FIG.
In the example of the extraction region ER1 shown in FIG. 1, m1, n1, m2, n2
Is the extraction area offset data. The extraction area offset data is also given from the first control unit 202 to the second control unit 212.

In step S31, the coded data included in the compressed image data Dc (i) of the i-th band is read by the separation circuit 204 by one bit. As described above, the separation circuit 204 uses the AC decoding circuit 206.
Is set to give encoded data to. The 1-bit encoded data is transferred from the separation circuit 204 to the AC decoding circuit 2
06, and also to the buffer 216 via the bit counter 214, where it is stored.
The bit counter 214 is a circuit that accumulates the number of bits of the encoded data supplied from the separation circuit 204.

In step S32, the AC decoding circuit 206
Is the Huffman code table 122 for AC coefficients (FIG. 14)
With reference to, the coded data read in step S31 and the code word of the Huffman code table 122 are compared. If the encoded data read in step S31 does not match any code word in the Huffman code table 122,
The process returns from step S33 to step S31 to read the encoded data by adding another bit. If the coded data matches any codeword in the Huffman code table 122 in step S32 (that is, if the coded data is decoded), steps S33 to S34.
Move to.

When the AC decoding circuit 206 decodes the coded data corresponding to one AC coefficient in step S33, the AC decoding circuit 206 gives the counter 210 a count-up signal CA. Counter 210
Counts the number of pulses of the count-up signal CA to substantially count the number of pixel blocks in the compressed image data Dc (i) along each scanning line,
The count value CT is given to the second controller 212. The counter 210 is initialized to 0 by the second controller 212 at the beginning of the extraction process of the compressed image data.

The second controller 212 controls the bit counter 2
A decoding end signal SF generated when all the AC coefficients of each pixel block have been decoded is given to 14. The bit counter 214 receives the decoded AC according to the signal SF.
The number of bits Nb of the encoded data is notified to the second control unit 212, and the number of bits Nb inside the bit counter 214 is reset to 0. In this way, the bit counter 214 sequentially outputs the bit number Nb of the encoded data of the AC coefficient for each pixel block to the second controller.
The second control unit 212 stores the given number of bits Nb in an internal register (not shown).

In step S34, the second controller 212 determines whether or not the pixel block being processed is a pixel block in the extraction area ER1. When the following expressions (6) and (7) are satisfied, the pixel block is in the extraction region ER1. n1 ≦ INT [{MOD (CT, N · K) −1} / K] + 1 ≦ n2 (6) m1 ≦ INT [(CT-1) / (N · K)] + 1 ≦ m2 (7)

Here, the function "INT" indicates an operation for rounding down the decimal point of the numerical value in the parentheses, and the function "MOD" is an operation for taking the remainder obtained by dividing the first value in the parenthesis by the second value. Show. CT is the count value of the counter 210, N is the number of pixel blocks in the main scanning direction y of the entire image IP (N in FIG. 18).
= 7), K is the number of conversion coefficients in the compressed image data of the i-th band (K = 7 in the second band B2), n1
Is the minimum value of the pixel block coordinates in the main scanning direction y of the extraction region ER1, and m1 and m2 are the minimum and maximum values of the pixel block coordinates of the extraction region ER1 in the sub scanning direction x.

When the pixel block being processed is the pixel block in the extraction region ER1, the AC coefficient coded data CFa is transferred from the buffer 216 to the memory 46 and stored therein in step S35. Second control unit 2
12 outputs the coded data CFa of the pixel block, that is, the data of the bit number Nb stored in the buffer 216, to the memory 46 based on the bit number Nb of the coded data CFa.

On the other hand, the pixel block being processed is the extraction area E.
If it is not a pixel block in R1, the process goes from step S34 to step S36 to clear the AC coefficient coded data CFa having the number of bits Nb stored in the buffer 216, and then proceeds to the next step S37.

In step S37, the second controller 212
Judges whether or not the decoding of the AC coefficient for one pixel block is completed, and if not completed, returns to step S31 and instructs the first control unit 202 to repeat the above processing. In this embodiment, the number K of AC coefficients for one pixel block in the second to fourth bands is as shown in FIG.
As shown in FIG. 3, the values are 7, 18, and 36, respectively, and these values are given to the second controller 212 in advance. The second controller 212 determines the number K of these AC coefficients.
It is determined whether or not the decoding of the AC coefficient for one pixel block is completed based on.

In step S38, it is determined whether the pixel block being processed is the final block in the extraction area ER1. The final block is a pixel block PB2 whose coordinate value (m, n) is equal to the maximum value (m2, n2) of the extraction region ER1 (see FIG. 18). When the following expressions (8) and (9) are satisfied, it is the final block in the extraction area ER1. INT [{MOD (CT, NK) -1} / K] + 1 = n2 (8) (6) INT [(CT-1) / (NK)] + 1 = m2 (9) Final If it is not a block, the process returns to step S31 and the processes of steps S31 to S38 are repeated.

When the coded data up to the final block of the extraction area ER1 has been extracted, the extracted compressed image data is edited by the CPU 42 and the image data supplying device 40 causes the image retrieving device 1 to operate in step S39.
Is transferred to 0.

FIG. 21 shows the extracted compressed image data ED.
It is explanatory drawing which shows c. As shown in FIG. 21, in the header of the extracted compressed image data EDc, in addition to the file name, band division data (data indicating the position coordinates of DCT coefficients included in the band), band number, and offset. Contains data. The variable length encoded data CFa follows the header. Image retrieval device 1
0 is the decompression processing unit 1 for the extracted compressed image data EDc.
The image expanded in 8 and the image quality of the extraction regions ER1 and ER2 improved is displayed on the CRT 36. Note that the extracted compressed image data EDc may be stored in a magnetic disk or the like in step 39 of FIG.

As described above, in this embodiment, only the compressed image data of the designated area is extracted from the compressed image data of the designated band.
Further, it is possible to reduce the amount of compressed image data that is the target of decompression processing. Therefore, there is an advantage that the image search processing can be performed at high speed. Further, when extracting the compressed image data, it is detected that the encoded data in the compressed image data matches one of the code words in the Huffman code table, and the delimiter between the encoded data is identified. It is possible to extract a part of the compressed image data that has been variable-length coded by the Huffman code.

E. Modification Note that the present invention is not limited to the above embodiment,
The present invention can be implemented in various modes without departing from the spirit of the invention, and the following modifications are possible, for example.

(1) The present invention is applicable not only to image data composed of YMCK image components, but also to expanding a part of compressed image data composed of BGR components and CIE color system components. .. It is also applicable to image data containing only one color image component. In general, the present invention can be applied to decompress multivalued compressed image data.

(2) In the above embodiment, the image data is compressed by encoding the image data by the DCT transform and the Huffman encoding. However, the present invention is generally a compressed image compressed by the sequence conversion and the entropy encoding. Applicable when extracting data. As the sequence transformation, for example, orthogonal transformation such as Fourier transformation or Hadamard transformation, vector quantization coding, or the like can be used. Further, arithmetic coding or the like can be used as the entropy coding.

(3) In the above embodiment, the compressed image data of the first band includes one DC coefficient and two AC coefficients, but the band division may be changed arbitrarily. For example, the first band may include only the DC coefficient. Further, the number of bands is not limited to four, and generally the conversion coefficient may be divided into a plurality of bands (hierarchies).

(4) In the above embodiment, the JPEG s-s progressive coding is adopted as the coding method, but the present invention is also applied to the JPEG s-a progressive coding. It is possible. Moreover, it is also applicable to image data compressed by another progressive coding method. As other progressive coding methods, “improvement of coding efficiency of hierarchical adaptive discrete cosine transform coding method”, Zheng Shige et al., Journal of Japan Society of Printing, Vol. 28, No. 3, 1991, pp. 33-33. 42, Laplacian Pyramid Encoding Method, Journal of the Television Society, Vol. 44, No. 2, 153 to 161, 1990.
There is a progressive coding method for binary images (international standard coding method for binary images (JBIG)).

(5) The range of the extraction area ER1 is CRT3
The operator operates the mouse 26 on the reduced image displayed in 6.
, But instead of this, the extraction area ER
The operator determines the width (pixel number) of 1 by the keyboard 28
You may make it input using. (6) As the image display device, another image output device such as a color printer may be used instead of the CRT 36.

(7) In the above embodiment, a part of the compressed image data is extracted from the non-interleaved compressed image data, but the present invention can be applied to the extraction of the compressed image data generated according to the interleave. In this case, the second controller 212 changes the number of pieces of decoded data, which is used as a criterion for determining whether the coded data to be processed is in the extraction area, according to the interleaving method. You can leave it.

[0064]

As described above, according to the invention described in claim 1, after displaying the whole image by decompressing the compressed image data of the first layer, the extraction area is designated and processed. Since the compressed image data corresponding to the extraction area is extracted from the compressed image data of the target layer and expanded, and the image is updated, the image quality of the extraction area can be improved. Therefore, it is possible to obtain an image of a desired image quality without decompressing the compressed image data of the entire screen for each layer.

According to the invention described in claim 2,
Based on the number of data (transform coefficients) obtained by decoding the entropy coded data, entropy coded data corresponding to the pixel block included in the extraction region is extracted, and the entropy coded data of the extracted extraction region ( That is, since the compressed image data) is expanded and the image is updated, the image quality of the extraction region can be improved. Therefore,
There is an effect that an image having a desired image quality can be obtained without expanding the compressed image data of the entire screen for each layer.

[Brief description of drawings]

FIG. 1 is an explanatory diagram showing an example of conversion coefficient band division.

FIG. 2 is a block diagram showing an image processing system for searching compressed image data by applying an embodiment of the present invention.

FIG. 3 is a block diagram showing the configuration of an encoding unit provided in a compression processing unit 48.

FIG. 4 is an explanatory diagram showing pixel blocks in an image.

FIG. 5 is a transformation coefficient F (p, p, obtained by DCT transformation.
Explanatory drawing which shows q).

FIG. 6 is an explanatory diagram showing an example of a quantization table used for quantization.

FIG. 7 is a block diagram showing the function of a DC coefficient encoding unit 78.

FIG. 8 is an explanatory diagram showing the contents of data processing in a DC coefficient encoding unit 78.

FIG. 9 is an explanatory diagram showing a grouping table.

FIG. 10 is an explanatory diagram showing a one-dimensional Huffman code table for DC coefficients.

11 is a block diagram showing the function of an AC coefficient coding unit 80. FIG.

FIG. 12 is an explanatory diagram showing a route of zigzag scanning.

FIG. 13 is an explanatory diagram showing DCT transform coefficients arranged in 8 rows and 8 columns and AC coefficients rearranged by a zigzag scan.

FIG. 14 is an explanatory diagram showing a two-dimensional Huffman code table for AC coefficients.

FIG. 15 is an explanatory diagram showing the structure of compressed image data Dc.

FIG. 16 is a diagram showing an image to be processed in the embodiment.

FIG. 17 is a flowchart showing the procedure of a search process for compressed image data.

FIG. 18 is an explanatory diagram showing the entire area and the extraction area of the image IP.

FIG. 19 is a block diagram showing circuit elements related to extraction of compressed image data.

FIG. 20 is a flowchart showing a detailed procedure of compressed image data extraction processing.

FIG. 21 is an explanatory diagram showing a configuration of compressed image data EDc created by extraction processing.

[Explanation of symbols]

 10 image retrieval device 12 CPU 14 bus line 16 memory 18 expansion processing unit 20 smoothing processing unit 22 magnetic disk 24 input interface 26 mouse 28 keyboard 30 communication interface 32 frame memory 34 D / A converter 36 CRT 38 image input device 40 image data supply device 42 CPU 44 bus line 46 memory 48 compression processing unit 50 compressed image extraction unit 58 interface 60 communication interface 62 magneto-optical disk 64 magnetic disk 72 DCT conversion unit 74 quantization unit 76 Huffman encoding unit 78 DC Coefficient coding unit 80 AC coefficient coding unit 82 Compressed data editing unit 102 Block delay unit 104 Adder 106 Grouping processing unit 108 One-dimensional Huffman coding unit 110 Huffman code table 112 Zigzag Can unit 114 Judgment unit 116 Run length counter 118 Grouping unit 120 Two-dimensional Huffman coding unit 122 Huffman code table 204 Separation circuit 206 AC decoding circuit 208 DC decoding circuit 210 Counter 214 bit counter 216 Buffer Aa Additional bit data Ad Additional bit data B1 to B4 band CA count-up signal CFa AC coefficient coded data CFd DC coefficient coded data Dc compressed image data Do original image data ER1 extraction area EDc compressed image data after extraction F DCT conversion coefficient Ha Huffman coded data Hd Huffman coded data IP All image NNNN Run length count PB Pixel block R1 Image quality improvement area SSSS Group number m Sub scanning direction pixel block coordinates n Main scanning direction pixel block Target x sub-scanning direction pixel coordinate y main scanning direction pixel coordinates

Claims (2)

[Claims] 1. A method of decompressing compressed image data composed of a plurality of hierarchical data, comprising: (a) displaying the entire image by decompressing the compressed image data of the first layer. (B) specifying the next layer to be the target of the extraction process, and designating the extraction region from which the compressed image data of the layer to be processed is to be extracted in the displayed whole image, (c) From the compressed image data of the layer to be processed,
Extracting the compressed image data corresponding to the extraction area, and (d) updating the image of the extraction area by decompressing the compressed image data of the processing target layer extracted in the step (c), A step of displaying the entire compressed image, and a method for decompressing compressed image data, comprising: 2. The image data is subjected to series conversion for each pixel block in the entire image along a scanning line in a predetermined direction, the conversion coefficient data obtained by the series conversion is divided into a plurality of layers, and each layer is A method of extracting a part of the compressed image data from the compressed image data created by entropy-encoding the transform coefficient data for each time and decompressing the compressed image data. A step of displaying the entire image based on the image data, and (b) specifying the next layer to be the target of the extraction processing, and extracting the compressed image data of the layer to be processed in the displayed entire image. (C) sequentially decoding the entropy-encoded data of the compressed image data of the layer to be processed and the number of decoded data Accumulating and extracting entropy-encoded data corresponding to the pixel blocks included in the extraction region based on the number of the decoded data; and (d) the entropy extracted in the step (c). Updating the image in the extraction area by decompressing the encoded data and displaying the entire updated image, the decompressing method of compressed image data.