JPH06176082A - Device and method for retrieving image - Google Patents

Abstract

PURPOSE:To provide an image retrieval method/device which can reduce the memory capacity of the data on the partial images, etc., used for register and retrieval of document images and furthermore can store even such document images that have unregistered formats after sorting these images for each for mat. CONSTITUTION:The format of an input document image is recognized (102) and collated with the registered formats (103). If no coincidence is confirmed between both formats, a new format is registered (106). In this case, the segmenting position of the document image of a relevant format is supplied (107) and stored. When a request is received for retrieval of the document image, the image is segmented based on the segmenting position.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image retrieval apparatus and method for registering and retrieving document images.

[0002]

2. Description of the Related Art A system for registering and retrieving document images is
For example, it is described in JP-A-63-118876 and JP-A-1-134671.

In Japanese Patent Laid-Open No. 63-118876, a document image is input, a plurality of partial images are cut out in accordance with a predetermined format registered in advance, the cut-out partial images are stored in a file, and a predetermined search code is set. A method of storing the partial image in the file in association with the storage address of the partial image is described. By this method, a document image and partial images cut out from the document image are automatically
Since it is possible to create a table in which search codes corresponding to each partial image are associated, by entering the search code at the time of search, the partial images can be read and a guidance screen can be easily created and displayed. A desired document image can be easily searched by specifying the item.

Japanese Laid-Open Patent Publication No. 1-134671 discloses an image processing apparatus for extracting an image index area from range designation information provided in a document image. The image index is a partial image displayed at the time of document search. Range designation information for designating an area used as an image index is provided for each document image, and when a document is read, the image index is extracted based on the range designation information at the time of searching. At the time of document search, a plurality of image indexes are displayed, and the user visually searches for the target document from among them.

[0005]

DISCLOSURE OF THE INVENTION Problems to be Solved by the Invention
In the technique described in Japanese Patent No. 8876, no consideration is given to changing the format. Therefore, when a document image having a format other than the predetermined format registered in advance is input, there is a problem that the document image cannot be registered. Further, in JP-A-63-118876 and JP-A-1-134671, since a partial image is cut out at the time of registration, there is a problem that the amount of image data such as a partial image increases. .

An object of the present invention is to register / search a document image in an image search apparatus and method applied to a search system for registering a partial image in a document image as a key and searching the document image based on the key. It is an object of the present invention to provide an image search apparatus and method capable of reducing the storage capacity of image data such as partial images.

Another object of the present invention is to provide an image retrieval apparatus and method capable of classifying and storing a document image even if the document image has an unregistered format. is there.

[0008]

In order to achieve the above object, the present invention stores a document image input by an image input means in a document image storage means, and stores the document image in the document image storage means at the time of document retrieval. In the image retrieval apparatus for displaying a list of cut-out images cut out from each of the document images displayed on the image display means, reading out the document image corresponding to the specified cut-out image and displaying the read-out document image on the image display means. Format storing means for storing the format, collating means for recognizing the format of the document image input by the image inputting means, and collating with the format stored in the format storing means, and the format storing means by the collating means. If the same format does not exist in, the format of the input document image is Stores in Tsu preparative storage means, and caused to enter the cut-out position of the document image, characterized in that a format registering means for storing in the position storage means cut the cut-out position.

When retrieving a document image, the document image stored in the document image storage means is cut out in accordance with the cutout position stored in the cutout position storage means, and the cutout images are displayed in a list.

According to the present invention, in an image retrieval apparatus for retrieving a plurality of document images of the same format stored in the document image storage means, means for reading out and displaying an arbitrary document image from the document image storage means. A cutout position designating unit that designates a cutout position by the displayed document image, and a unit that cuts out the document images stored in the document image storage unit according to the designated cutout position and displays the cutout images in a list. A means for reading out and displaying a document image corresponding to a specified cut-out image among the list-displayed is provided.

Further, according to the present invention, the input document image is stored, and at the time of document search, a list of cut-out images cut out from each of the stored document images is displayed.
In an image retrieval method for reading out and displaying a document image corresponding to a clipped image designated from among them, a step of inputting a document image, a step of recognizing a format of the input document image, and a recognized format is already formatted. If the recognized format is not yet stored in the format file, the format of the input document image is stored in the format file and the document image A step of inputting a cutout position and storing the cutout position.

In the image retrieval method for retrieving a plurality of document images of the same format stored in the document image storage means, a step of reading out and displaying an arbitrary document image from the document image storage means is displayed. A clipping position designating step of designating a clipping position by the document image, a step of clipping the document images stored in the document image storage means according to the designated clipping position and displaying the clipped images in a list, And reading and displaying a document image corresponding to the cut-out image designated from the inside.

[0013]

In the present invention, first, the operator inputs a document image using the image input means. The format is recognized from the received document image. The recognized format is
It will be checked against the registered format. As a result of the collation, if the same format as the document image format is already registered, the format is not registered and the document image is registered in the file managed by the format.

If the format of the input document image is not registered yet, that format is newly registered. When registering a format, the operator inputs the cutout position and stores the input cutout position. As a result, the cutout position corresponding to the format is stored, so that the document images of the same format can be cut out at the same cutout position.

After that, when the operator requests the retrieval of the document image, the document image is first clipped according to the stored clipping position, and the clipped image is displayed as a list on the image display means. The operator refers to the cut-out images displayed in a list and searches for a target document. When the operator specifies one cutout image using a pointing device or the like, a document image corresponding to the cutout image is displayed.

As described above, at the time of document registration, it is possible to save the trouble of file classification and key input and reduce the volume of image data such as partial images.

[0017]

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

FIG. 1 is a flow chart showing a registration method of an image search method according to an embodiment of the present invention. Input / output is illustrated at the same time as the document image registration procedure.

In FIG. 1, reference numeral 41 is a document image. 1
Is an image input device for inputting the document image 41. A display device 2 displays a document image 41. 3 is a keyboard. Reference numeral 4 is a pointing device such as a mouse. Reference numeral 11 is an image registration program.

A format file 20 stores the format of the document image. A cut-out position table 30 stores the cut-out position designated by the pointing device 4. A document image file 40 stores the registered document image 41. An image registration address table 50 stores document image registration addresses.

Reference numerals 101 to 110 are processing steps of the image registration program. First, the document image 41 is received from the image input device 1 (step 101). The format of the received document image 41 is recognized (step 102),
The format is checked against the format already registered in the format file 20 (step 103). As described above, the technique of collating images is a known technique. When the input document image 41 is input with a tilt, for example, the tilt is corrected and collation is performed. The technique for automatically correcting the inclination is a known technique.

If the same format is found in the format file 20 in step 103, it is judged that the format is already registered (step 10
4) Execute step 109 for the file name registered in association with the format file 20. If the same format is not found in the format file 20 in step 103, step 104
Then, the process proceeds to step 105 via the display to display the document image 41 on the display device 2 (step 105).

Next, the input of the file name is accepted from the keyboard 3, and the format associated with the accepted file name is newly registered in the format file 20 (step 106). Then, the partial image to be cut out from the pointing device 4 is accepted (step 10
7) The clipping position of the designated partial image is registered in the clipping position table 30 (step 108).

Next, the input document image 41 is registered in the document image file 40 (step 109), and the document image registration address is stored in the image registration address table 50 managed by the file name input when the format is registered. Is written (step 110), and the process ends.

FIG. 2 is a flow chart showing a procedure for retrieving a document using the clipping position already registered in this embodiment. FIG. 3 is a flowchart in the case of performing a search using an unregistered cutout position. In these figures, the input / output is shown as well as the search procedure. The same numbering as in FIG. 1 indicates the same.

In FIG. 2, 12 is an image retrieval program, and 60 is a cutout image display position 6 illustrated in FIG.
2 is a document image search table that stores 1 and a document image registration address 51 in association with each other.

Reference numerals 111 to 126 are processing steps of the image search program. First, the search method is accepted from the keyboard 3 using the search menu (step 111). Here, a search method is specified in which a partial image of each document is displayed in a list, the user specifies the partial image of the target document from the list, and the document corresponding to the specified partial image is displayed. To do. The cutout position of the partial image to be displayed is specified in steps 107 and 108 of FIG. 1 and registered in the cutout position table 30, but the partial image is cut out and displayed at the registered cutout position, or is registered. Whether to cut out and display the partial image at a new new cutting position can be selected from the menu in step 111.

Next, the file name to be searched is accepted from the keyboard 3 (step 112). Step 111
Then, if a search method using an unregistered cutout position is specified as the search method, the process proceeds to step 131 in FIG. When the method of searching using the registered cutout position is designated (step 113), the cutout position is read from the cutout position table 30 managed by the file name accepted in step 112 (step 114).

Next, the document image registration address registered in the image registration address table 50 managed by the file name accepted in step 112 is read (step 115). The document image 41 illustrated in FIG. 5 is read from the document image file 40, and the image is cut out according to the cutout position (step 116). Then, the cutout image 5 is displayed at the cutout image pasting position 61 of the display device 2 (step 117). The cut-out image pasting position 61 is stored in the document image search table 60 in association with the document image registration address.

Next, if the document image 41 to be cut out exists in the image registration address table 50 (step 11).
9) and determine whether display is possible on the display device 2 (step 12
0), if display is possible, step 115 is repeated.
As a result, a list of cutout images that can be displayed on the display device 2 is displayed.

If there is no document image 41 to be cut out in the image registration address table 50 in step 119, or if the cutout image cannot be displayed in step 120,
Input from the keyboard 3 or the pointing device 4 is accepted (step 121). If the input specifies a clipped image (step 122), the document image search table 60 of FIG. 5 is searched at the specified position and the document image registration address is read (step 125). Then, the document image 41 is read from the document image file 40 by the document image registration address, the document image 41 is displayed on the display device 2 (step 126), and the process is ended.

If the input at step 122 does not specify the cutout image but the scrolling is specified (step 123), scrolling is performed (step 12).
4) Repeat step 115. If the input is not scroll designation (step 123), step 121 is repeated.

Referring to FIG. 3, description will be given of a processing procedure in the case of specifying an unregistered cutout position and performing a search.
In FIG. 3, 131 and 132 are processing steps for newly specifying the cutout position. First, the top document image 41 of the search target file is read from the document image file 40 and displayed on the display device 2 (step 131). next,
The cutting-out position is accepted by the pointing device 4 (step 132), and step 115 of FIG. 2 is executed. As a result, in the processing after step 115, the document image is cut out at the newly specified cutout position,
A list will be displayed and a search will be performed.

FIG. 4 is a format registration of the image retrieval method of this embodiment (mainly steps 105 to 110 in FIG. 1).
The flow of data in FIG. In FIG. 4, 7 is a storage device of the image processing apparatus. Reference numeral 21 is a format of the document image 41 composed of key lines and the like. Reference numeral 22 is a file name for managing the format. 31 is a document image 41
The cutout position of is shown. Reference numeral 51 is a document image registration address.

First, the document image 41 input by the image input device 1 is displayed on the display device 2. This is step 1 in Figure 1.
05. Next, use the keyboard 3 to enter the file name 22
Accept. For example, assume that “application form” is registered as the file name 22. File name 22'Application form '
The format 21 of the document image 41 is registered in the format file 20 in association with. The file name is accepted and the format is registered in step 106 of FIG.

After that, the operator designates the cut-out position (the upper left end (x coordinate, y coordinate) and the lower right end (x coordinate, y coordinate)) with the pointing device 4. This is according to step 107 in FIG. In the example of FIG. 4, the range written in the “name” field of the application form is specified as the cutout position.

Next, (x coordinate, y coordinate) at the upper left end
And (x-coordinate, y-coordinate) in the lower right corner, file name 2
2 Store in the cutout position table 30 managed by the'application form '. In this example, the (x coordinate,
Since the y coordinate is (60, 80) and the lower right end (x coordinate, y coordinate) is (200, 100), these coordinates are stored as the cutout position 31. The x coordinate is the distance from the left end of the document image 41, and the y coordinate is the distance from the upper end of the document image 41. The unit of both the x coordinate and the y coordinate is a dot.

After that, the document image 41 is registered in the document image file 40, and the document image registration address 51'0126 'is registered in the image registration address table 50 managed by the file name 22'application form'. Registration of the document image and registration of the address are performed in steps 109 and 110 of FIG.

FIG. 5 shows a data flow when a clipped image is displayed (mainly from step 112 in FIG. 2) in the document search of the image search method of this embodiment. In FIG. 5, reference numeral 5 is a partial image obtained by cutting out the document image 41 according to the cutout position 31. 6 is a CPU. Reference numeral 61 is a cut-out image pasting position for storing the position where the cut-out image 5 is displayed on the display device 2.

File name 2 to be searched with the keyboard 3
2 When the “application form” is specified, the cutout position 31 ′ (60, 80) (200, 100) ′ managed by the file name 22 “application form” from the cutout position table 30.
Read out. This is due to steps 112 and 114 in FIG.

Next, the document image registration address 51'4F2D 'managed by the file name 22'application form' is read from the image registration address table 50, and the document image 41 is read from the document image file 40. This is shown in Figure 2.
Step 115, 116.

Next, the CPU 6 cuts out the image in accordance with the cut-out position 31 and displays it on the cut-out image pasting position 61 of the display device 2. This is step 117 of FIG.
by. The y coordinate of the upper end of the cut image pasting position is
(Lower limit of pasting position pasted before) + 1. The y-coordinate of the lower end of the cut-out image pasting position can be calculated by (upper limit) + (height of cut-out image 5 (y-coordinate of lower right end of cut-out position-y-coordinate of upper left end))-1.

Finally, the cut-out image pasting position 61'6
1-80 'and the document image registration address 51'4F2D' are associated and registered in the document image search table 60. This is according to step 118 of FIG.

"Hitachi Taro", "as cut-out image 5"
Hitachi Hanako ',' Taro Kawasaki ', and'Hanako Chiba' are illustrated. As the document image registration address 51 managed by the file name 22'application form ',' 0126 ','
3452 ',' 0A3B ', and' 4F2D 'are illustrated. As the cut-out image pasting position 61 of the document image search table 60, "1 to 20", "21 to 40", and "41" are set.
~ 60 ', '61 -80', document image registration address 51
As' 0126 ',' 3452 ',' 0A3B ','
4F2D 'is illustrated.

FIG. 6 shows a process of selecting one of the cut-out images displayed as described above in the document search of the image search method of this embodiment until the document is displayed (mainly in FIG. 2).
The flow of data in steps 121, 125, and 126) of FIG. In FIG. 6, 62 is a pointing device 4.
Is the y coordinate of the point indicated by.

For example, it is assumed that the operator wants to retrieve the document image 41 called "Hitachi Taro". As a child, the operator
The cut-out image 5'Hitachi Taro 'of the display device 2 is designated by the pointing device 4. This is step 1 in FIG.
21.

Next, the y coordinate of the designated point 62'15 '
The document image search table 60 is searched with to read the document image registration address 51'0126 '. This is according to step 125 of FIG. Registered document image registration address 51 '
The document image file 40 is searched at 0126 ′, and the document image 41 corresponding to the cut-out image 5′Hitachi Taro ′ is read out and displayed on the display device 2. This is step 1 in FIG.
According to 26.

As the cut-out image 5, "Hitachi Taro", "
Hitachi Hanako ',' Taro Kawasaki ', and'Hanako Chiba' are illustrated. As the cut-out image pasting position 61 of the document image search table 60, "1 to 20", "21 to 4"
0 ', '41 -60', '61 -80 ', and' 0126 ',' 3452 ',' 0 as the document image registration address 51.
A3B 'and' 4F2D 'are illustrated.

FIG. 7 shows a device configuration of an image search device for realizing the above-described image search method.

In FIG. 7, 8 is an image processing apparatus.
First, the image registration program 11 is executed, and the document image 41 shown in FIG. 4, for example, is input from the image input device 1. The format 21 is recognized from the input document image 41 and collated with the format 21 registered in the format file 20 stored in the storage device 7. If the format 21 is already registered, the document image 41 is registered in the document image file 40, and the image registration address table 50 managed by the file name 22 registered in association with the registered format 21 is registered. register.

When the operator requests a document image search, the image processing apparatus 8 executes the image search program 12. When the operator specifies the file name 22 using the keyboard 3, the file name 22 is extracted from the cutout position table 30.
The cut-out position 31 managed by is read out.

Next, the document image registration address 51 managed by the designated file name 22 is read from the image registration address table 50, the document image 41 is read from the document image file 40, and the cut-out image 5 is created by the CPU 6. Then, it is displayed on the display device 2. The cut-out image pasting position 61 and the document image registration address 51 are associated with each other and stored in the document image search table 60. When the cutout image 5 desired to be searched by the pointing device 4 is shown, the document image search table 60 is searched from the y coordinate 62 of the point indicated by the pointing device 4, the document image registration address 51 is read, and the document image 41 is read from the document image file 40. Read out and display on the display device 2.

[0053]

According to the present invention, document images having the same format are automatically classified at the time of document registration, and the cutout position of the partial image is stored at the time of registration, and the partial image is cut out and displayed at the time of search. Therefore, it is possible to save the labor of key input and reduce the volume of image data such as partial images.

[Brief description of drawings]

FIG. 1 is a flowchart showing a registration method of an image search method according to an embodiment of the present invention.

FIG. 2 is a flowchart of a case where a document is searched using a clipping position that is already registered in the image search method according to the embodiment.

FIG. 3 is a flowchart of a case where an unregistered cutout position is designated and a search is performed in the image search method according to the embodiment.

FIG. 4 is a diagram showing a flow of format registration data in the image retrieval method of the embodiment.

FIG. 5 is a diagram showing a data flow when a clipped image is displayed in a document search by the image search method according to the embodiment.

FIG. 6 is a diagram showing the data in the document search of the image search method of the embodiment.
It is a figure which shows the flow of data.

FIG. 7 is a diagram showing a device configuration of an image search device for implementing the image search method of the embodiment.

[Explanation of symbols]

1 ... Image input device, 2 ... Display device, 3 ... Keyboard, 4
... pointing device, 11 ... image registration program, 20 ... format file, 30 ... clipping position table, 40 ... document image file, 41 ... document image,
50 ... Image registration address table, 101-110 ... Processing steps.

 ─────────────────────────────────────────────────── ─── Continuation of front page (72) Inventor Tsutomu Tashiro 1099, Ozenji, Aso-ku, Kawasaki City, Kanagawa Prefecture Hitachi Ltd. Systems Development Laboratory

Claims (6)

[Claims] 1. A document image input by an image input means is stored in a document image storage means, and a clipped image cut out from each document image stored in the document image storage means is displayed on an image display means at the time of document retrieval. In an image retrieval apparatus for displaying a list, reading out a document image corresponding to a specified clipped image from the list and displaying it on an image display unit, a format storage unit for storing the format of the document image and the image input unit for inputting the format. If the same format does not exist in the format storage means by the matching means for recognizing the format of the document image and the format stored in the format storage means and the format stored in the format storage means by the matching means, The format is stored in the format storage means and the document image is cut out. Image retrieval apparatus characterized by comprising a format registering means for storing in the position storage means cut the cut-out position and brought inputting position. 2. The image search device according to claim 1,
When searching for a document image, the document image stored in the document image storage means is cut out in accordance with the cutout position stored in the cutout position storage means, and the cutout images are displayed in a list. Search device. 3. An image retrieving apparatus for retrieving a plurality of document images of the same format stored in a document image storage means, and means for reading out and displaying an arbitrary document image from the document image storage means. A cutout position designating unit for designating a cutout position by the document image, a unit for cutting out the document images stored in the document image storage unit according to the specified cutout position, and displaying the cutout images as a list, An image retrieving apparatus, comprising means for reading and displaying a document image corresponding to a clipped image designated from the inside. 4. An input document image is stored, and at the time of document retrieval, a list of cut-out images cut out from each of the stored document images is displayed, and a document corresponding to the cut-out image designated from the list is displayed. In an image retrieval method for reading and displaying images, a step of inputting a document image, a step of recognizing a format of the input document image, and a collating step of collating whether or not the recognized format is already stored in a format file. If the recognized format is not already stored in the format file, the format of the input document image is stored in the format file, and the cutout position of the document image is input and the cutout position is stored. An image retrieval method comprising steps and . 5. The image search method according to claim 1, wherein
An image retrieval method characterized in that, when retrieving a document image, the stored document image is clipped according to the stored clipping position and a list of the clipped images is displayed. 6. An image retrieval method for retrieving a plurality of document images of the same format stored in a document image storage means, a step of reading out and displaying an arbitrary document image from the document image storage means, A clipping position specifying step of designating a clipping position by the document image, a step of clipping the document images stored in the document image storage means according to the designated clipping position and displaying the clipped images in a list, And a step of reading out and displaying a document image corresponding to a clipped image designated from the inside, and an image search method.