JP2606560B2 - Document image storage device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a document image storage method in a character recognition device, and more particularly to a document image storage method in a character recognition device for electronically managing a document.

[0002]

2. Description of the Related Art Conventionally, in order to manage a large number of existing document images, digitizing the document images has been performed. However, in this case, not only the data of the document image is simply compressed and the keyword is assigned, but also by analyzing the image and encoding the contents of the text area by character recognition, more efficient storage and retrieval can be performed. Analysis of document images is considered to be an important technology in the future for realizing such a function because it may be possible.

In a conventional document image storage system of a character recognition apparatus, a text area and a chart area which are constituent elements of a document are separated from an image obtained by inputting the document from an image scanner, and the inclusion relation and arrangement of the areas are separated. The relationship is stored as layout information, and particularly in the case of a text area, character recognition is performed separately for each character area.

As a first example of a document image storage system in such a conventional character recognition device, Japanese Patent Application No. 62-292 is disclosed.
07 “document image analysis method”. According to this method,
There is an advantage that any document can be coded, regardless of the format of the document image, such as vertical writing, horizontal writing, and multiple columns.

FIG. 6 is a block diagram showing a first example of a document image storage system in such a conventional character recognition device. A document image memory 61 shown in FIG. 6 stores image information obtained by digitizing a document image. In addition, the area dividing unit 6
2 divides the image information in the document image memory 61 from the global area to the local area while maintaining the arrangement relation of the vertical and horizontal relations, and based on the result of the area division, the document is vertically divided. Whether writing or horizontal writing is determined, and the result is stored in the vertical and horizontal information storage unit 63.

If the partial area obtained by the area dividing section 62 is a text area, the data of the text area is sent to the character separating section 64. The character separation unit 64
Line regions constituting the text region are extracted, and region information for each character constituting the line region is extracted and sequentially stored in the structured data storage unit 65.

Therefore, the area search section 67 reads out the condition relating to the arrangement relation of the area to be extracted, which is stored in the area definition storage section 66 in advance, and stores the readout condition in the structured data storage section 65 in accordance with the condition. Searching for structured data indicating the arrangement structure between regions in the document image. Further, the area search section 67 reads out the area satisfying the condition from the structured data storage section 65 and stores the area in the extraction result storage section 68 in a predetermined order.

As described above, in the first example of the document image storage method in the conventional character recognition device, a document image can be divided into constituent elements and each piece of layout information can be stored.

As a second example of a document image storage system in a conventional character recognition device, image data of a form is read, and characters included in the image data of the form are recognized and corrected. There is a method of creating a data file in which image data is linked and managing a form image and a character code in the form image in association with each other.

As a second example of the document image storage method in such a conventional character recognition device, Japanese Patent Application No. Hei 2-3937 is disclosed.
4 There is a “form processing device”. According to this apparatus, there is an advantage that a form image can be searched for by a character string written in the form image.

As shown in these examples, in the conventional document image storage system of the character recognition device, characters are coded and stored, but the type and font size of the character font constituting the document, etc. No character attributes have been obtained. That is, in the conventional techniques, only the positional relationship of the characters and the layout information of the document are stored.

[0012]

In the document image storage method in the above-described conventional character recognition apparatus, layout information and character codes are held, but the type and font size of character fonts are identified by current technology. This has not been realized due to the difficulty, and has the disadvantage that such information has been lost.

[0013] However, the document contains high-level information such as indicating emphasis items using a character font or a font size. In order to retain more advanced information such as this emphasis matter when digitizing a document, other information such as character codes obtained by character recognition and document layout information indicating the positional relationship and inclusion relationship between text areas and figure / table areas are included. In addition, it is necessary to store character attributes in a document such as font information.

SUMMARY OF THE INVENTION An object of the present invention is to provide a document image storage method in a character recognition device capable of storing character images in association with character attributes in order to retain advanced document information using character fonts. is there.

[0015]

According to a first aspect of the present invention, there is provided a character image storage system for storing a document image which is digitized image information data;
By extracting components such as text and diagrams from the document image stored in the document image storage unit, and determining whether the components are a text region or a diagram region,
A layout that outputs layout information including the inclusion relationship between the components and the top, bottom, left and right arrangement relationships, and outputs the contents of the character region divided into single characters as a character image when the component is a text region. Analysis means; layout information storage means for storing layout information obtained by the layout analysis means; and character recognition for recognizing a character image obtained by the layout analysis means and outputting the recognition result as a character code. Means, a character code correction means having a function of correcting and replacing a part of the character code obtained by the character recognition means with an appropriate character code, and a character image obtained by the layout analysis means, A character image storing the character code obtained after the correction by the character code correction means in association with the character code It is configured by including a 憶 means.

The document image storage method in the character recognition device according to the second invention is characterized in that, in addition to the constituent elements of the document image storage method in the character recognition device according to the first invention, By extracting constituent elements such as sentences and figures from the document image stored in the document image storing means of the first invention, it is determined whether the constituent elements are a text area or a chart area. When the constituent element is a chart area, it has a function of outputting the contents of the constituent element as a chart image, and stores a chart image output by the layout analyzing means. The layout information stored in the layout information storage unit is read, and a change is made to the arrangement relationship, and the character image stored in the character image storage unit of the first invention and Nde read a chart images stored in the table image storage means is configured to include a layout editing means for outputting performing rearrangement in accordance with the modified arrangement relationship.

On the other hand, the document image storage system in the character recognition apparatus of the third invention stores a document image which is digitized image information data, and extracts components such as sentences and figures from the document image. By determining whether the constituent element is a text area or a chart area, layout information including an inclusion relation between the constituent elements and an upper, lower, left and right arrangement relation is output, and the constituent element is the text area. In the case of, the content of the character area divided into one character unit is output as a character image, the obtained layout information is stored, and the obtained character image is recognized. Output, correct and replace a part of the obtained character code with an appropriate character code, and obtain the character code obtained by correcting the obtained character image. Association with storing includes that.

Further, in the document image storage system in the character recognition apparatus according to the fourth invention, a document image which is digitized image information data is stored, and constituent elements such as sentences and figures are extracted from the document image. By determining whether the constituent element is a text area or a chart area, layout information including an inclusion relation between the constituent elements and an upper, lower, left and right arrangement relation is output, and the constituent element is the text area. In the case of, the contents of the character area divided in units of one character are output as a character image, and when the constituent element is the chart area, the contents of the constituent element are output as a chart image and are obtained. The obtained layout information is stored, the obtained character image is recognized, the recognition result is output as a character code, and a part of the obtained character code is appropriately used. Correcting and replacing the character code, storing the obtained character image in association with the character code obtained after the correction, storing the obtained chart image, reading the layout information, and arranging the layout. Reading the character image and the chart image with a change to the relationship, performing rearrangement according to the changed layout relationship, and outputting the result.

[0019]

Next, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing a first embodiment of a document image storage system in the character recognition device of the present invention. First, the document image storage means 10 shown in FIG. 1 stores a document image which is digitized image information data.

The layout analysis unit 11 reads the document image 100 obtained from the document image storage unit 10 and divides the document image 100 into components such as a text area and a chart area. It outputs layout information 101 of an arrangement relationship of up, down, left, and right. If the layout analysis unit 11 determines that the component is a text area, it extracts a line area wrapped in the text area,
A character area for each character included in the line area is extracted, the layout information 101 is output, and the character area is output as a character image 102.

On the other hand, the layout information storage means 12 stores the layout information 10 obtained from the layout analysis means 11.
1 is stored. The character recognizing means 13 recognizes the character image 102 obtained from the layout analyzing means 11 and outputs a character code 103 as a recognition result.
Then, the character code correcting means 14
Is replaced with a character code to be corrected, and the replaced character code 104 is output.

Therefore, the character image storing means 15 stores the character image 102 obtained from the layout analyzing means 11 and the character code 104 obtained from the character code correcting means 14 in association with each other.

Next, the operation of the first embodiment will be described. First, the digitized document image stored in the document image storage unit 10 is sent to the layout analysis unit 11. The layout analysis unit 11 divides the document into components of a text area and a chart area, and stores the inclusion relation and the arrangement relation in the layout information storage means 12.

Then, for a region determined to be a text region, for example, a line region forming a text region and a character unit region forming a line region are extracted by division processing. In addition, each positional relationship and arrangement relationship,
The layout information is stored in the layout information storage unit 12 as layout information.

FIG. 2 is a view showing an example of a character image cut out by the layout analysis means 11. In FIG. This allows
For example, the text area shown in FIG.
The image is separated into the character areas shown in FIG.

On the other hand, FIG.
FIG. 3 is a diagram showing an example of data stored in a storage device. The inclusion relation and arrangement relation of the character images are indicated by, for example, hierarchical layout information as shown in FIG. Here, the pointer 30 points to a text area, the pointer 31 points to a line, the pointer 32 points to a character, and the physical structure of the layout is expressed hierarchically. Further, a character code recognized by the character recognition unit 13 is added to the character image.

Further, the character code correcting means 14 corrects the character code. Here, a manual correction function of the erroneously recognized character is included. The character image is stored in the character image storage unit 15 together with the character code corrected by the character code correction unit 14.

FIG. 4 is a diagram showing an example of data stored in the character image storage means 15. Thereby, for example, as shown in FIG. 4, the character image and the character code are stored in association with each other. As described above, storage in which character codes are associated with character images is achieved.

FIG. 5 is a block diagram showing a second embodiment of the document image storage system in the character recognition device of the present invention.
In the second embodiment of FIG. 5, in addition to the components of the first embodiment, the layout analysis unit 16 stores the chart image storage unit 1
7, a layout editing unit 18 is connected to the layout information storage unit 12, the character image storage unit 15, and the diagram image storage unit 17.

Next, the operation of the second embodiment will be described. First, the layout analysis unit 16 reads the document image 100 obtained from the document image storage unit 10, divides the document into components such as a text region and a diagram region, and includes the inclusion relationship of the components and the top-bottom / left-right arrangement relationship. Is output.

When the layout analyzing means 16 determines that the constituent element is a text area, it extracts a line area included in the text area, and extracts a character area in character units included in the line area. The character image 102 is output, and the layout information 101 is output. Further, when it is determined that the component is a diagram or a table, the diagram image 105 is output.

On the other hand, the chart image storage means 17 stores the chart image 105. The layout editing unit 18 stores the document layout information 10 from the layout information storage unit 12.
6, the layout information is changed, and the character image 107 stored in the character image storage unit 15 and the chart image 108 stored in the chart image storage unit 17 are rearranged and edited according to the changed arrangement information. The result 109 is output.

[0033]

As described above, when the document image storage method in the character recognition device of the present invention is used, for example,
Once an existing document stored in a database or the like has been digitized and retrieved, it can be reproduced in the same way as the original document, including attribute information such as character emphasis in the document. is there.

Further, the document image storage method in the character recognition device of the present invention has an effect that it is possible to perform an output with a changed layout, such as printing a vertically written document horizontally while maintaining the character attributes. ing.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a first embodiment of a document image storage system in a character recognition device of the present invention.

FIG. 2 is a diagram showing an example of a character image cut out by a layout analysis unit 11;

FIG. 3 is a diagram illustrating an example of data stored in a layout information storage unit 12;

FIG. 4 is a diagram illustrating an example of data stored in a character image storage unit 15;

FIG. 5 is a block diagram showing a second embodiment of the document image storage system in the character recognition device of the present invention.

FIG. 6 is a block diagram showing a first example of a document image storage method in a conventional character recognition device.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 10 Document image storage means 11 Layout analysis means 12 Layout information storage means 13 Character recognition means 14 Character code correction means 15 Character image storage means 16 Layout analysis means 17 Chart image storage means 18 Layout editing means 30, 31, 32 Pointer 61 Document image Memory 62 Area division unit 63 Vertical and horizontal information storage unit 64 Character separation unit 65 Structured data storage unit 66 Area definition storage unit 67 Area search unit 68 Extraction result storage unit

 ──────────────────────────────────────────────────続 き Continuation of front page (56) References JP-A-5-108866 (JP, A) JP-A-3-161886 (JP, A) JP-A-3-131992 (JP, A)

Claims (4)

(57) [Claims] 1. A document image storage means for storing a document image which is digitized image information data, and constituent elements such as sentences and figures are extracted from the document image stored in the document image storage means. By judging whether a component is a text area or a chart area, output layout information including the inclusion relation between the components and the top, bottom, left and right arrangement relations, and when the component is a text area A layout analysis unit that outputs the contents of a character area divided into single character units as a character image, a layout information storage unit that stores layout information obtained by the layout analysis unit, and a layout analysis unit that obtains the layout information. Character recognition means for recognizing a character image and outputting the recognition result as a character code; and A character code correcting unit having a function of correcting and replacing a part of the obtained character code with an appropriate character code; and obtaining a character image obtained by the layout analyzing unit after correcting the character image by the character code correcting unit. Character image storage means for storing the associated character codes in association with each other,
A document image storage device comprising: Wherein said layout analysis means, said extracting components of text and diagrams, such as from the stored document image in the document image storage in means, whether the components are either figure area is the text area Determining, when the component is a chart area, having a function of outputting the contents of the component as a chart image, and storing a chart image output by the layout analyzing means; Reading the layout information stored in the layout information storage means and making changes to the layout relationship, reading the character image stored in the text image storage means and the chart image stored in the chart image storage means, and changing the layout relationship. document image according to claim 1, wherein the layout editing means for outputting performing rearrangement in accordance, characterized in that it comprises to 憶apparatus. 3. A document image, which is digitized image information data, is stored, and constituent elements such as text and charts are extracted from the document image to determine whether the constituent element is a text area or a chart area. By judging, the layout information including the intentional inclusion relation of the constituent elements and the top, bottom, left and right arrangement relations is output, and when the constituent element is the text area, the character area divided into one character unit is output. The content is output as a character image, the obtained layout information is stored, and the obtained character image is recognized, and the recognition result is output as a character code, and a part of the obtained character code is output. Correcting and replacing with an appropriate character code, and storing the obtained character image in association with the obtained character code.
Document image storage device . 4. A document image, which is digitized image information data, is stored, and constituent elements such as text and charts are extracted from the document image to determine whether the constituent element is a text area or a chart area. By judging, the layout information including the inclusion relation between the constituent elements and the vertical, horizontal, and horizontal arrangement relations is output, and when the constituent element is the text area, the contents of the character area divided into one character unit Is output as a character image, and when the component is the chart area, outputs the content of the component as a chart image, stores the obtained layout information, and obtains the obtained character image. By recognizing, the recognition result is output as a character code, a part of the obtained character code is corrected and replaced with an appropriate character code, and the obtained character image is corrected. The character codes obtained later are stored in association with each other, the obtained chart images are stored, the layout information is read, and the arrangement relationship is changed, and the character images and the chart images are read. A document image storage device , which performs rearrangement according to the changed arrangement relation and outputs the result.