JPH11203305A - Method for processing document picture and recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To automatically extract a part (key area) useful to the grasping a content from a document picture, and to narrow down the number of extraction into a proper number. SOLUTION: An area dividing layout information extraction means 102 divides an inputted document picture into a character area and an element such as a graphic, chart, or ruled line, and extracts layout information such as the kind of a multiple column or the multiple column. A line extracting means 103 extracts a line from a character area, and a font identifying means 104 identifies a font by each line unit. A key area discriminating means 105 extracts a key area directly indicating a document content based on the feature of the layout expression, checks the total number of the extracted key areas, and reduces and narrow-down the number into a proper number when the number is beyond a threshold value.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a document image processing method and a recording medium for automatically extracting a region useful for grasping contents from a document image.

[0002]

2. Description of the Related Art In recent years, opportunities for distributing document images have dramatically increased due to improvements in the capabilities of computers and digital devices and in the capabilities of data storage devices such as hard disks. However, a document image has a large amount of data, which has been a bottleneck when a processing speed is required or when image data is exchanged through a network.

In general, when browsing a large number of document images, it is not necessary to view all the images in the state of the stored high-quality images. Is often sufficient. Of course, processing can be performed much faster when a keyword search or the like can be performed than when an image is handled. However, it is difficult to enter appropriate keywords. It is even more difficult if the stored image content has never been seen.

Therefore, in an optical filing apparatus or the like, the reduced image is held as an index image separately from the original image, and the index image is first presented to the user to reduce the amount of data to be processed. I have. However, if you simply shrink the entire image, it is difficult to read the characters in the image even if you can grasp the whole feeling,
In particular, when there are many documents having similar layouts, it is difficult to select a desired document.

To solve such a problem, there is a document processing apparatus described in Japanese Patent Application Laid-Open No. Hei 5-342326. This apparatus divides a document image into regions, assigns logical identifiers to the divided elements according to a logical model, extracts only necessary logical elements using the identifiers as keys, and rearranges them so that they are easy to see. In addition, partly OC
By using R, sorting of extracted information is also performed.

[0006]

However, in the above-described document processing method, it is necessary to comprehend the layout configuration and the logical configuration of the document image input in advance and create a corresponding model. The document is structured according to the model,
If the region dividing unit is perfect, processing can be performed with high accuracy. However, in many cases, processing is not so accurate, and the application range of the model is limited. In addition, since considerable skill is required to create a logical model, it is very difficult to extract a partial image (key area) to grasp the contents of a new document group in the above-described processing method.

[0007] As another method for solving the above problem,
For example, there are Japanese Patent Application Laid-Open No. 5-242142 and Japanese Patent Application No. 8-110808 previously proposed by the present applicant. These methods do not need to comprehend the layout configuration and the logical configuration and create a corresponding model, and are more flexible than the method disclosed in Japanese Patent Laid-Open No. Hei 5-342326.

That is, in the method for summarizing a document without decoding a document image described in Japanese Patent Application Laid-Open No. 5-242142, a linguistic standard (such as "important" or "significant") is used. The importance, that is, the semantically important image unit is determined based on the word), the position in the document, and the morphological image characteristics (character style, character type, etc.). However, in this method, even if it is possible to determine the “importance” of each image unit (area), it does not control how much the whole image becomes. By using frequency information, it is possible to control the total number of important equivalent areas, but in Japanese sentences, even if the frequency of occurrence is low, it often corresponds to a "key area". Is difficult to control.

In the method proposed in Japanese Patent Application No. 8-110808, a document image is divided into a plurality of elements, and layout characteristics of each of the divided elements (emphasized portions different from the main text). In this method, it is determined whether or not the element is an area that expresses the contents of the document, and the area is extracted as a partial image. In this method, the “importance” of the area is also determined. Even if you can, it does not control how much it will be in the whole image.

The present invention has been made in view of the above circumstances, and an object of the present invention is to automatically extract a portion (key area) useful for grasping contents from a document image.
Another object of the present invention is to provide a document image processing method and a recording medium in which the number of extractions can be reduced to an appropriate number.

[0011]

In order to achieve the above object, according to the first aspect of the present invention, a document image is divided into a plurality of elements, and each of the divided elements is laid out on the basis of a layout expression characteristic. What is claimed is: 1. A document image processing method for determining whether an element is an area (hereinafter, referred to as a key area) that briefly represents the content of the document image, comprising: determining the number of the determined key areas;
It is characterized in that the number of key areas is reduced to an optimum number.

The invention according to claim 2 is characterized in that when the number of the determined key areas is equal to or larger than a predetermined threshold, the number of the key areas is reduced.

According to a third aspect of the present invention, when the ratio of the number of determined key areas to the total number of lines in a page is equal to or greater than a predetermined threshold, the number of key areas is reduced. It is characterized by.

[0014] The invention according to claim 4 is characterized in that priorities are assigned to the features in the layout expression, and the key regions are sequentially reduced from key regions determined based on the features having the lower priority.

According to the fifth aspect of the present invention, the same label is assigned to key areas having the same characteristics, the number of key areas for each label is counted, and the key area is reduced according to the count for each label. It is characterized by.

According to the sixth aspect of the present invention, the document image is divided into a plurality of elements, and based on the layout representation characteristics of each of the divided elements, the elements clearly represent the contents of the document image (hereinafter referred to as "areas"). , A key area), a column configuration of the entire page of the document image is determined, and the key area is determined according to the determined column configuration. The feature is to select a feature in the layout expression to be used.

According to the present invention, the document image is divided into a plurality of elements, and based on the layout representation characteristics of each of the divided elements, the elements clearly indicate the contents of the document image (hereinafter referred to as "areas"). A key area), the number of the determined key areas is checked, and when it is necessary to reduce the number of key areas, the layout expression used for the determination is used. It is characterized in that the key area is determined again by using a feature excluding one or more of the above features.

According to the eighth aspect of the present invention, a function of dividing a document image into a plurality of elements, and an area where the elements express the contents of the document image based on the layout representation characteristics of each of the divided elements. (Hereinafter referred to as a key area) a computer-readable recording medium storing a program for causing a computer to realize a function of determining whether or not the key area is a key area, the function of checking the number of the determined key areas; It is a computer-readable recording medium that records a program for causing a computer to realize a function of reducing the number of key areas to an optimum number.

[0019]

DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be specifically described below with reference to the drawings. <Embodiment 1> FIG. 1 shows the structure of an embodiment of the present invention. In the figure, 101 is an image input means, 102 is an area division / layout information extraction means for dividing a document image into elements and extracting layout information such as columns, and 103 is a line extraction for extracting a line in each character area. Means 104, font identification means, 105 key area discriminating means for discriminating an area (key area) which expresses the contents of the document image from information extracted by the area division / layout information extracting means 102 and line extracting means 103, Reference numeral 106 denotes a data storage unit for storing input images and various information during processing, 107 a control unit for controlling the whole, and 108 a data communication path.

FIG. 2 shows a processing flowchart of the present invention. Hereinafter, the present invention will be described with reference to FIG. First, a document image is obtained by the image input means 101 (step 20).
1). The image input means may be a scanner or a facsimile, or may be a means for obtaining an image from another device via a network.

Next, the area dividing / layout information extracting means 102 divides the input document image into character areas and elements such as figures, tables, and ruled lines, and at the same time, lays out layout information such as column types and columns. Extract (step 202). As such an area dividing method, for example, Japanese Patent Laid-Open No. 6-200
A known technique described in JP-A-92-92 may be used. As a method for extracting layout information, for example, a known technique described in Japanese Patent Application Laid-Open No. 9-44594 may be used. The extracted area has, as attributes, the type of element (character area, table, etc.), its position (circumscribed rectangle of the area), and the like, and the entire image has layout information such as a column type and a column dividing line.

The line extracting means 103 extracts a line from the extracted character area (step 203). For example, the IEICE paper “Marginal distribution, cotton density,
The technique described in “Area Segmentation of Document Image Using Bounding Rectangular Feature” (Akiyama et al., August 1986, Vol. J69-DNo. 8) may be used.

Further, the font identifying means 104 identifies the font for each line or for each character included in each line (step 204). As the font identification method, for example, a method described in Japanese Patent Application Laid-Open No. 6-208649 may be used.

When layout information such as areas, line information, and font information are extracted, the key area determining means 105 determines a key area (step 205). Here, the key area is a part useful for grasping the contents of the document.
Generally, such parts are emphasized in some way. For example, processing such as using a large character or an emphasis type font, making it independent from other parts, and enclosing it with a frame is performed. Here, the earlier application for Japanese Patent Application Hei 8
Utilizing a document image processing method proposed in Jpn. Pat.

FIG. 3 is a detailed processing flowchart of the key area determination processing. First, a key area is extracted by checking the emphasis processing on the layout expression for each unit (here, a row) (step 301).

That is, in the key area extraction processing proposed in the above-mentioned application, first, the character size characteristics of each line (uppercase lines,
A middle character line, a normal character line) is detected, and then a font feature (whether or not the font is emphasized based on black pixel density or the like) is detected, and a title portion (a large character size and an independent region) is detected. Perform detection. Subsequently, a subheading portion (for example, an area in which the number of lines is less than a predetermined number in a middle character line) is detected,
Bibliographic information is detected, and finally, if a box exists,
The text in the box is often a separate article and is useful for understanding the content, so several lines in the box are extracted from the beginning.

For example, the "title area", "font emphasis area", "subtitle area and line", "bibliographic information", "enclosing frame top line", etc., obtained by the above processing are key areas.

After extracting key areas by checking the emphasis processing on the layout expression for each unit (here, row), the total number of extracted key areas is checked (step 30).
2). By the way, since the key area aims at the effect of summarizing the document image, if too many lines are extracted as the key area, the effect is reduced by half. Therefore, in the present invention, the total number of extracted key areas is checked, and the key areas are reduced when a predetermined condition is satisfied. This condition includes, for example, a case where the total number of extractions is equal to or greater than a predetermined threshold, or a case where the ratio of the number of extracted key areas to the total number of lines in the page is equal to or greater than a predetermined threshold. These conditions may be OR conditions or AND conditions.

The key area reduction processing (step 303) is performed as follows. FIG. 4 is a detailed processing flowchart of step 303. First, being determined as a key area indicates that there is some emphasis processing in expression. In other words, this is the case where an emphasis font is used, or the case where it is determined that the font is a subheading based on the indent or the character size. Therefore, in the processing of step 301, the same label is assigned to the key areas determined to have the same characteristics. FIG. 5 shows a state of labeling. Then, the number of key areas is counted for each label.

When it is determined that the number of key areas is too large as a whole, a reduction target is determined. As a determination method, there are a method of (1) a priority order of a predetermined emphasis feature and a method of (2) a reduction from a large number of the same labels.

In the method (1), in the example of FIG. 5, the priority of title equivalent> subtitle equivalent> font emphasis is determined in advance, so that the key area of only the font emphasis having the lowest priority is first reduced. If the number of key areas is still large, the key area corresponding to the subheading is reduced. Also, when there are a plurality of emphasized features (for example, equivalent to a subheading and font emphasis), the priority is equal to or higher than that of a single high-priority feature.

In the method of (2), in the example of FIG. 5, the labels having the largest number of identical labels (= font emphasis) are reduced, and this is repeated until the number of key areas becomes sufficient.

<Embodiment 2> The configuration of this embodiment is the same as that of FIG. 1, and the overall processing flow is also shown in FIG.
And different.

In step 202, for example,
-45594 (A plurality of small areas including a character string are extracted from a document image, a blank portion or a ruled line is detected from the plurality of small regions, and one step is performed based on the detected blank portion or the ruled line. The column structure of the entire page is determined by using a column type discriminating method for determining a column type including a set, a plurality of columns, and a free column. Here, for example, the whole is 1
An example will be described in which a column is determined. A one-column document includes a notification sentence and the like, and the indentation tends to be frequently used in the notification sentence. In addition, an emphasis type Gothic font may be used for a part that is not particularly important.

Therefore, when there is a deep indent as shown in FIG. 6 in a one-column system, in the process of step 205, the indent feature is not used for the key area determination, or the weight as the feature is reduced, or Processing such as not using font information.

As described above, by changing the weight of the feature used for the key area determination depending on the column configuration of the entire page, a more optimal key area is extracted.

<Embodiment 3> The configuration of Embodiment 3 is the same as that of FIG. 1, and the overall processing flow is also shown in FIG. The difference lies in the processing content of the key area determination processing in step 205.

FIG. 7 is a detailed processing flowchart of step 205 according to the third embodiment. As in Example 1,
First, a key area is extracted (step 701), and the first embodiment is performed.
Similarly to the above, the number is determined (step 702),
If it is necessary to reduce the number of areas (when there are too many corresponding areas), the key area is extracted again without using one or more features for key area extraction (step 70).
3). Steps 702 and 703 may be performed recursively.

<Embodiment 4> The present invention is not limited to the above-described embodiment, but can be realized by software. When the present invention is realized by software,
As shown in FIG. 8, a computer system including a CPU, a memory, a display device, a hard disk, a keyboard, a CD-ROM drive, a mouse, and the like is prepared. CD-
A computer-readable recording medium such as a ROM stores a program for implementing the document image processing function and the processing procedure of the present invention. The document image to be processed is stored, for example, on a hard disk or the like. Then, the CPU reads a program for realizing the above-described processing function and processing procedure from the recording medium, extracts key areas from a document image read from a hard disk or the like, and, if the number of key areas is excessive, the number of key areas. Is reduced to an optimal number, and the result is output to a display or the like by changing colors.

[0040]

As described above, claims 1 and 2,
According to the inventions described in the third and eighth aspects, when an area useful for grasping the contents of a document image is excessively extracted, the number of the areas is reduced, so that an optimal amount can be obtained.

According to the fourth and fifth aspects of the present invention, since an object to be reduced is taken into consideration, an appropriate amount of area useful for grasping the contents of the document image can be obtained.

According to the sixth aspect of the present invention, an area useful for grasping the contents of the document image can be obtained in an appropriate amount according to the column structure of the entire page.

According to the present invention, when an area useful for grasping the contents of the document image is excessively extracted, the key area is discriminated again by using different features.
It is possible to obtain an optimum amount of an area that expresses the contents.

[Brief description of the drawings]

FIG. 1 shows a configuration of an embodiment of the present invention.

FIG. 2 shows a processing flowchart of the present invention.

FIG. 3 is a detailed processing flowchart of a key area determination process.

FIG. 4 is a detailed processing flowchart of step 303 (key area reduction processing).

FIG. 5 shows a state of labeling for a key area.

FIG. 6 shows an example of deep indentation.

FIG. 7 is a detailed processing flowchart of step 205 (key area determination processing) in another embodiment.

FIG. 8 shows a configuration example when the present invention is realized by software.

[Explanation of symbols]

 Reference Signs List 101 Image input means 102 Area division / layout information extraction means 103 Line extraction means 104 Font identification means 105 Key area determination means 106 Data storage unit 107 Control unit 108 Data communication path

Claims (8)

[Claims] 1. A document image is divided into a plurality of elements, and based on layout characteristics of each of the divided elements, the elements are areas (hereinafter, referred to as key areas) that briefly represent the contents of the document image. What is claimed is: 1. A document image processing method, comprising: determining the number of determined key areas; and reducing the number of key areas to an optimal number. 2. The document image processing method according to claim 1, wherein when the number of the determined key areas is equal to or more than a predetermined threshold, the number of the key areas is reduced. 3. The method according to claim 2, wherein the number of key areas is reduced when a ratio of the number of determined key areas to a total number of lines in a page is equal to or greater than a predetermined threshold. 1. The document image processing method according to 1. 4. The document image processing method according to claim 1, wherein priorities are assigned to the features in the layout expression, and key regions determined based on features with lower priorities are sequentially reduced. 5. The method according to claim 1, wherein the same label is assigned to key areas having the same characteristics, the number of key areas for each label is counted, and the key area is reduced according to the count for each label. 1. The document image processing method according to 1. 6. A document image is divided into a plurality of elements, and based on layout characteristics of each of the divided elements, the elements are areas (hereinafter, referred to as key areas) that briefly represent the contents of the document image. A document image processing method for determining whether or not a column structure of an entire page of the document image is determined, and according to the determined column structure, a layout expression used for determining the key area is determined. A document image processing method characterized by selecting a feature. 7. A document image is divided into a plurality of elements, and based on layout characteristics of each of the divided elements, the elements are areas (hereinafter referred to as key areas) that express the contents of the document image. A document image processing method for determining whether or not the number of key areas determined, and when it is necessary to reduce the number of key areas, among the features in the layout expression used for the determination, A document image processing method characterized by re-determining a key area using a feature excluding one or a plurality of features. 8. A function of dividing a document image into a plurality of elements, and an area where the elements express the contents of the document image based on layout characteristics of the respective elements (hereinafter referred to as a key area). A computer-readable recording medium on which a program for causing a computer to realize a function of determining whether or not a key area is determined is provided. A computer-readable recording medium on which a program for causing a computer to realize the function of reducing the number is recorded.