JP2851894B2 - Display control method for image retrieval device - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image database search device, and more particularly to a display control method for a search device using hierarchically encoded images.

[Conventional technology]

When searching and retrieving a target image from a conventional image database, an outline of the image is recognized using a thinned image having a small number of pixels (low-resolution image), and a human recognizes whether or not the image is the target image. He used the technique of searching.

[Problems to be solved by the invention]

However, in the above conventional example, even if the outline of the image can be grasped, fine information such as character / line drawing information in the image cannot be decoded because of the coarse resolution, and is ignored as information, resulting in a decrease in search capability. . In addition, if the character / line drawing information is displayed at a resolution that allows humans to determine the contents sufficiently, the thinning rate cannot be so large, and therefore it must be displayed at a high resolution. There was a problem in terms of search response. An object of the present invention is to prevent such inconveniences.

[Means for solving the problem]

In order to solve the above problem, the present invention encodes image data of one screen hierarchically at different resolutions for each of a plurality of areas, and adds a delimiter code indicating a delimiter of data of each layer. In a display control method of an image search device capable of decoding and displaying an image at a plurality of resolutions from a stored storage medium, a first method for determining whether an area to be decoded from an image to be read from the storage medium is a high-resolution display area or not. And decoding the image of the high-resolution display area,
The first code for counting the delimiter code until a count value corresponding to a predetermined high resolution is reached, and decoding a predetermined high resolution image based on the encoded data of the high resolution display area.
When decoding an image in an area different from the high-resolution display area, the delimiter code is counted until a count value corresponding to a predetermined low resolution is obtained, and based on the encoded data in the different area. A display control method of an image search device includes a second decoding step of decoding a predetermined low-resolution image and a display step of displaying the images decoded in the first and second decoding steps.

(Operation)

The present invention enables high-speed search while maintaining a resolution sufficient for an operator to determine a desired image by the above configuration.

〔Example〕

 Hereinafter, embodiments of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram of an embodiment of an image retrieval apparatus to which the present invention is applied. In FIG. 1, reference numeral 1 denotes a control unit (CPU) for controlling the overall operation of the image retrieval apparatus, 2 a keyboard for inputting various commands for image inspection to the CPU 1, 3 a display for displaying images and characters, 4 Is a reader for photoelectrically reading a document image and inputting image data, 5 is a database capable of storing image data for a plurality of pages, 6 is a frame memory for storing at least one screen of data to be displayed on the display 3, and 9 is a frame memory. An encoding device 10 for encoding image data is a decoding device for decoding the encoded image data.

Now, the image is read from the reader 4 and the database 5 is read.
A description will be given starting from the case where the data is stored.

First, the image data input from the reader 4 is hierarchically encoded by an encoding device 9, and the database 5 is sequentially converted from low-resolution encoded data to high-resolution encoded data.
Is accumulated in

FIG. 2 is a block diagram showing a configuration example of the encoding device 9. 12, 14, 16 are frame memories, 13, 15 are reduction circuits, 1
7, 18, and 19 represent encoders. First, original image data output from the reader 4 and desired to be hierarchically encoded is stored in the frame memory 12. The original image data stored in the frame memory 12 is reduced by the reduction circuit 13 to an image of half the size in the vertical and horizontal directions and stored in the frame memory 14. The reduction method at this time may be, for example, a method of performing sub-sampling to 1 / n (here, n is 2) in both the vertical and horizontal directions. Alternatively, a method of performing sub-sampling after threshold value processing of the obtained value may be used, but the method is not limited thereto. Similarly, the half-size image data stored in the frame memory 14 is further reduced to a half-size image in the vertical and horizontal directions by the reduction circuit 15 and stored in the frame memory 16. Therefore, the frame memory 16 stores the image stored in the frame memory 12 vertically and horizontally.
A 1/4 size image is stored.

The encoder 17 performs entropy encoding of the image data stored in the frame memory 16 by using, for example, an arithmetic code technique, and outputs encoded data 101 of the first stage.
The encoder 18 entropy-encodes the image stored in the frame memory 14 with reference to the image stored in the frame memory 16, and outputs the second-stage encoded data 102. Similarly, in the encoder 19, the frame memory 14
The image stored in the frame memory 12 is entropy-encoded with reference to the image stored in the, and the encoded data 103 in the third stage is output. The image data hierarchically encoded in this way is sequentially recorded in the database in ascending order of resolution.

Here, entropy coding is taken as an example of a coding method, but other coding may be used. Further, it is possible not only to store hierarchically encoded image data in a database, but also to transmit it to another system through a communication line.

Also, the stage of the reduced image is up to 1/4 vertically and horizontally as an example, but if it can be recognized as an image, 1/8, 1/16, 1/32 ...
It is also possible to create low resolution images.

FIG. 3 shows an example of an image sample to be made into a database. It is assumed that one image (for example, A4 size) is divided into rectangular areas (called blocks) 1 to M as shown in the figure, and is coded on a block basis.

In the upper (head) position of the image, additional information of the image is written in characters, marks, illustrations, and the like. The additional information includes the date and time when the image was taken, the contents, the photographing conditions (eg, film type, shutter speed, exposure value, etc.), the photographer, the index (serial number and classification number) attached to the image, and the characteristics. Mark or illustration.

As the additional information, humans can use the information added for organizing and classifying a plurality of images as it is, and can save the trouble of inputting a lot of information from a keyboard which is conventionally used as an image database.

This additional information is written in a limited area of the block (for example, block 1) and is assumed to be written in another area (block 2).
) Contains a normal image (natural image or the like).

FIG. 4 shows a file configuration of encoded data when this image is encoded. The image for one page is hierarchically encoded for each block. For example, the lowest resolution coded data of block 1 (for example, 1/16) is stored in the file area 31a, the next 1/8 data is stored in 31b, and so on. In the image data of the next block 2, 1/16 data is stored in the area 32a, 1/8 data is stored in the area 32b, and 1/1 data of the block M is stored in the area 40e. Actually, such encoded data is variable-length data, and the memory is arranged one-dimensionally.
As shown in FIG.
This delimiter code 51 usually uses a code of several bytes that never appears in the data after encoding the image, and indicates the delimiter of each layer and each block.

The plurality of encoded image data configured in this way are stored in a database.

FIG. 6 shows a block diagram of the decoding apparatus. 20, 21, and 22 are decoders, and 23, 24, and 25 are frame memories. First, among the encoded data (hierarchically encoded signals) hierarchically encoded and stored for each block in the database, the encoded data 201 of the first stage (low resolution) is decoded by the decoder 20. Is input to the frame memory 23, where the result is decoded.
And outputs the decoded image data 301 in the first stage of the first block. To restore a higher resolution image in the first block, the following processing is performed.

The encoded data 20 of the second stage among the encoded data (the hierarchically encoded signals) read from the database 5
2 is input to the decoder 21. In the decoder 21, the coded data 202 is decoded with reference to the image data stored in the frame memory 23 at the time of the preceding first-stage decoding, and the result is stored in the frame memory 24. The image data 302 thus output is output. Further, the encoded data 203 of the third stage (high resolution) of the encoded data (the hierarchically encoded signal) read from the database is input to the decoder 22. In the decoder 22, the encoded data 203 is decoded with reference to the image data stored in the frame memory 24, and the result is stored in the frame memory 25 and output as the third-stage image data 303. The image data stored in the frame memory when decoded to the final stage in this way represents the original image before being encoded.

In this way, the block 1 is restored to a resolution at which characters, line drawings and the like can be decoded. In order to stop at an intermediate resolution and proceed to the next block, it is sufficient to skip the intermediate code data, and it is possible to count only the delimiter code and know the head of the next stripe.

Next, for the image portion (blocks 2 to M), only low-resolution image data may be extracted from the encoded data of each block and combined. In general, the file structure has a one-dimensional structure shown in FIG. 5 as described above, and reads data from low resolution to high resolution in each block shown in FIG.
When reading of image data of one block is completed, the same operation is performed in the next stripe. (Fig. 4 shows 42a to 4
Proceed to 2b. In order to extract only low-resolution data from the image portion, it is necessary to skip data at an intermediate resolution. The number of delimiters is counted, and the head of the next stripe is obtained. This is the same as in the case of the character / line drawing described above.

As such a delimiter code, it is possible to use different delimiter codes for each layer and for each block. At this time, it is only necessary to find the delimiter at the end of the block.

FIG. 7 is a block diagram of a circuit for finding this delimiter code at high speed. The encoded image data is sequentially sent to the shift register 60. The delimiter code data is stored in, for example, a register 62 or the like. The data of a predetermined bit length in the one-dimensional string packed in the shift register 60 and the delimiter code are compared with the comparator 61, and if they match, 1 is matched. If not, 0 is obtained from output 63.

In this way, the presence or absence of a delimiter code can be determined by sequentially sending encoded data to the shift register.

8 (A) and 8 (B) are flowcharts showing the above operation, where (A) shows encoding and (B) shows decoding.

In (A), encoding is performed hierarchically for each block. Assuming that the number of blocks M is Mmax and the number of coding layers N is Nmax, in step 101, M is initialized to 1;
First, a reduced image of each layer (only in the block) from N = 1 to Nmax is created, and the reduced image of each layer is stored in the memory (steps 103 to 105). Subsequently, the reduced image of the hierarchy Nmax is encoded (step 106). Next, the layer N is decremented by one, and the reduced image of the layer N is subjected to hierarchical coding with reference to the reduced image of the layer N + 1, and is repeated up to the layer 1 (steps 107 to 109). Then, this processing is repeated up to block Mmax (steps 110 and 111).

It should be noted that a delimiter code indicating a block delimiter is added after the encoded data of each block, and the encoded data is stored in the memory. Then, the image data of each layer of each block of one screen is read from the memory and stored in the database as shown in FIG. 4 (step 112). In addition, the location of the delimiter of the character portion is fixed (for example, the first block), and after the first block is encoded, a delimiter code is automatically added. Then, the operator may use a keyboard or the like to specify the location of the separation.

Next, the decoding operation will be described in (B). First, header information of a desired image is received from the database (step 121). As the header information, a block number (Mc) containing character information and the like is set. Subsequently, the first block M = 1 is set (step 122). It determines whether they match with a character information block number Mc in step 123, if I match, set the final stage N _P of decoding the N ₁ there so as to decode up to high resolution (N ₁ layer) (Step 125). Matching the N _P so as performing complex to lower resolution (N ₂ layer) unless set with N ₂ (step 124) (Note that it is N _2> N _1).

After the above setting, Nmax is set to N so that decoding starts from the last layer N = Nmax (step 12).
6). The first decoding is performed while reading the encoded data from the database until a delimiter code of the resolution is detected (steps 127 and 128). Then it is determined whether the hierarchy N meets the N _P (step 129), if not match counts down the hierarchy N by 1 (step
130), decrypt again to the next delimiter code. If the hierarchy N is N _P , it is determined in step 131 whether the hierarchy N is 1 or not. As a result, if they do not match, the next delimiter code is detected and the layer N is counted down (step 132). At this time, decoding is not performed. In this way, the detection of the delimiter code is performed until N = 1. That is, this portion is skipped, the image data is not decoded, and the head of the next block is detected. The above processing is performed until the block number M reaches Mmax.

Since from the above description is N ₁ <N ₂ as self-evident, the area of the blow stick number Mc that contains the text information in high resolution (up to N ₁ layer), the other for low resolution (N ₂ Decoding is performed). If N ₁ = 1, the character area is decoded to the maximum resolution. The block number Mc for providing the character area is not limited to one, but may be plural.

As described above, it is possible to change and display the resolution at the time of decoding for each block. Therefore, the character / line drawing portion can be displayed at a high resolution and the image portion can be displayed at a low resolution, and both the information of the character / graphic and the image can be used as a search means, thereby enabling a high-speed search.

In the above-described embodiment, the information position of the character and the line drawing is moved to the upper part. However, any position may be used. However, if this position changes for each image, it is necessary to have information indicating what block number the block comes from before the code data.

(Effect)

As described above, according to the present invention, when decoding to display an image hierarchically coded for each area, a count value corresponding to a predetermined high resolution or a count value corresponding to a predetermined low resolution is used. By counting the delimiter code until the value is reached and decoding images with different resolutions for each region, it is possible to reliably display images at different resolutions for each region with a simple configuration using the characteristics of the encoding method. It is possible to quickly confirm whether or not the image is a desired image.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of an image retrieval apparatus to which the present invention can be applied, FIG. 2 is a block diagram showing a configuration of an encoding apparatus, FIG. 3 is a view showing an image to be encoded, FIG. FIG. 5 is a diagram showing a storage state of encoded image information, FIG. 6 is a block diagram showing a configuration of a decoding device, FIG. 7 is a diagram showing a circuit for detecting a delimiter code, and FIG. 8) is a flowchart showing an encoding process, and FIG. 8 (B) is a flowchart showing a decoding process. 3 Display 5 Database 9 Encoding device 10 Decoding device 13, 15 Reduction circuit

──────────────────────────────────────────────────続 き Continuation of the front page (56) References JP-A-63-61554 (JP, A) JP-A-63-156476 (JP, A) JP-A-60-134357 (JP, A) JP-A 63-61357 109653 (JP, A) JP-A-63-76070 (JP, A) JP-A-56-14286 (JP, A) JP-A-2-3872 (JP, A) JP-A-63-290459 (JP, A) (58) Field surveyed (Int.Cl. ⁶ , DB name) G06F 17/30

Claims (1)

(57) [Claims] An image data for one screen is coded with a hierarchically different resolution for each of a plurality of areas, and a plurality of storage media are stored from a storage medium in which a delimiter code indicating a delimiter of data of each layer is added and stored. In a display control method of an image search device capable of decoding and displaying an image at a resolution, a first determination step of determining whether an area to be decoded of an image to be read from the storage medium is a high-resolution display area; When decoding the image of the display area, the delimiter code is counted until a count value corresponding to a predetermined high resolution is reached, and a predetermined high resolution image is decoded based on the encoded data of the high resolution display area. 1 decoding step, when decoding an image in an area different from the high resolution display area, counting the delimiter code until a count value corresponding to a predetermined low resolution is reached. A second decoding step of decoding a predetermined low-resolution image based on the coded data of the different areas, and a display step of displaying the images decoded in the first and second decoding steps. A display control method for an image search device, comprising: