JP2000067158A - Document image processing method and machine-readable recording medium where program allowing computer to implement document image processing method is recorded - Google Patents

Abstract

PROBLEM TO BE SOLVED: To correctly extract character areas from a document image even if characters belonging to different character areas are close to each other. SOLUTION: The document image processing method which corrects excessive integration of extracted character areas when performing character recognition and layout information acquisition as to character areas by extracting the character areas from a document image includes a step S201 for inputting the document image, a step S202 for reducing the inputted document image, extracting the circumscribed rectangles of connecting components of black pixels constituting characters, and extracting basic elements, a step S203 for classifying the basic elements by characters, tables, figures, and others, generating lines by integrating the character elements, and extracting the character areas by integrating the lines, a step S204 for extracting column setting information from the character areas, and a step S205 for correcting over-integrated character areas by referring to the positions of extracted columns.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention is used in a document image recognition system for recognizing an input document image, and in particular, when extracting a character region from a document image, corrects excessive integration of the extracted character regions. The present invention relates to a document image processing method and a machine-readable recording medium storing a program for causing a computer to execute the document image processing method.

[0002]

2. Description of the Related Art Conventionally, in order to input a document image, perform character recognition processing from the document image, and obtain various layout information, it is necessary to perform area extraction processing such as a character area, a graphic area, and other areas. There is.

[0003] As a reference technical document for performing such an area extraction process, "Printed Document Recognition System AutoRec"
o / 2 -Image process- "(Information Processing Society of Japan 47th
(The second half of 1993) National Convention 2-97 to 98) is disclosed. Here, in particular, a column boundary is detected, and character string elements are integrated into an area (this is referred to as a first conventional technique).

Japanese Patent Laid-Open No. 9-445 discloses a technique for detecting a column state of an entire page and extracting a character area accordingly.
No. 94, which discloses a "method of dividing a document image into regions and a method of determining a column type" (this is referred to as a second prior art).

[0005]

However, the prior art as described above has the following problems. First, in the first prior art, if the caption of a figure is close to the text in a newspaper or the like, excessive integration may occur. FIG. 7 shows a specific example. In FIG. 7, since the caption 704 is close to the text, it is erroneously extracted as one area in a form like 703.

In the method disclosed in the second prior art, when character elements are dense, such as in a newspaper, a clear column is not detected in an initial stage, and neighboring character elements are integrated. The character area will be extracted. That is, excessive integration occurs when approaching as shown in FIG.

Further, even in the method disclosed in the first prior art, adjacent character elements are grouped in an initial stage, so that an edge portion of a step cannot be detected.
Even in such a case, if the edge portion of the step is detected, a large number of pseudo edges other than the original edge are detected, and the extraction of the character area becomes complicated as a whole.

The present invention has been made in view of the above, and when extracting a character region from a document image, it is possible to correctly extract a character region even if characters belonging to different character regions are close to each other. With the goal.

[0009]

In order to achieve the above object, a document image processing method according to claim 1 extracts a character region from a document image, and performs character recognition and layout information for the character region. In a document image processing method for correcting over-integration of extracted character regions when performing acquisition, etc., an image input step of inputting a document image, and a document image input in the image input step are reduced to form characters. A basic element extraction step of extracting a circumscribed rectangle of a connected component of a black pixel to be extracted and extracting a basic element;
Generate a line by classifying it into others and integrating character elements.
A character region extraction step of integrating the lines to extract a character region, a column extraction process of extracting column information from the character region extracted in the character region extraction process, and a column extraction process of the column extraction process. And correcting an excessively integrated character area with reference to the position of the character area.

In the document image processing method according to the second aspect, in the document image processing method according to the first aspect, the area correction step may be performed by over-integration from the position of the column extracted in the column extraction step. Is detected, the character or character string element in the character area is integrated into the character area again with reference to the column position, and the degree of line consistency with the preset reference division position is evaluated. This is to correct the area.

In the document image processing method according to a third aspect, in the document image processing method according to the second aspect, the area correction step is performed when the area outside the reference division position is integrated with the character area again. First, a character region composed of a direction perpendicular to the reference line direction located at the position is generated, and then the character region is evaluated.

According to a fourth aspect of the present invention, there is provided a machine-readable recording medium on which a program for causing a computer to execute the document image processing method according to any one of the first to third aspects is recorded. is there.

[0013]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, a document image processing method and a machine readable recording medium storing a program for causing a computer to execute the document image processing method of the present invention will be described in detail with reference to the accompanying drawings.

(System Configuration) FIG. 1 is a block diagram showing the configuration of a system to which a document image processing method according to an embodiment of the present invention is applied. In the figure, 101 is CC
This is an image input unit for optically reading and inputting a document including a document as a binary image using a scanner or a facsimile device mounted on the D. The image input unit 10
As a specific means, a document image may be obtained from another device via a network (by a data communication unit 110 described later).

When an image is taken in via a network, for example, a browser function (WWW (w
and a wide-area data resource (URL which is one of access information on a WWW server) distributed on the Internet using an old wide web (a browser function).
(Uniform Resource Locato
r) (Hyper Text M)
arkup Language: WWW (world
wide web: browses data in a language such as a document description language for an Internet information service using hypertext).

Reference numeral 102 denotes an image reduction unit for reducing the image input by the image input unit 101; 103, a basic element extraction unit for extracting character elements from the image reduced by the image reduction unit 102; A character region extraction unit 105 to be extracted is a column detection unit 106 that detects a column from the character region extracted by the character region extraction unit 104 and the like.
Is a character area correction unit that executes a character area correction process described later, 107 is a control unit that controls the entire system in accordance with a predetermined control program, and 108 is various data such as input document data and layout information. Is a data storage unit for storing data, 109 is a data communication path, 1
10 is, for example, TCP (Transmission C).
control Protocol / IP (Inter)
This is a data communication unit configured to perform network communication according to a “net protocol”.

(System Operation) Next, the document image processing operation in the system configured as described above will be described. FIG. 2 is a flowchart showing a procedure of the document image processing method according to the embodiment of the present invention. First, a target document image is input by the image input unit 101 (S20).
1). Further, the input document image is converted to an image
The size is reduced by 02 (S202).

In this case, for example, by OR-compressing 8.times.8 pixels into one pixel, black pixels constituting the character are connected in the character string direction. The circumscribed rectangle of this connected component is extracted and used as a basic element. This basic element may be a single character or a part of a character string, or a table alone may be one large basic element.

Subsequently, the basic elements are classified into characters, tables, figures, and the like, and the character elements are integrated to generate a line, and the lines are integrated to extract an area (S203). In addition,
The flow of this processing is described in detail in
No. 4 proposes this.

Subsequently, column information is extracted (S20).
4). This column will be described with reference to FIGS. The length of the extracted character area along the line direction is "area width", and the length in the direction perpendicular to the "area width" is "area height".
And Here, a histogram of “region width” is created for a region whose “region height” is equal to or larger than a preset threshold.

That is, in the case of FIG. 3, 301 to 304 are character areas having a height equal to or higher than the threshold value. This histogram is shown in FIG. In the histogram of FIG. 4, when there is a class having a high frequency with respect to the total frequency, the class is defined as “step width”. In FIG. 4, a class having a width of 90 to 120 corresponds to this. In FIG. 3, reference numerals 301 to 303 correspond to this.

At this time, a limit may be set to a range where the "step width" can be taken. For example, in the example shown in FIG.
If 0 is within the range, even if the frequency of the class of 30 to 60 is high, it is not adopted as the step width, and the step width is undecided finally. In the method disclosed in Japanese Patent Application Laid-Open No. 9-44594, a consistent column state is detected from the distribution of character elements, and the detected column information may be used.

If the column information has been extracted, the character area is corrected using the column information (S205). The details of the character area correction processing are shown in the flowchart of FIG.

In FIG. 5, first, an area matching the estimated step width is detected (S501). In the example shown in FIG.
Corresponds to. Subsequently, a candidate area that should belong to the same row as the area detected in step S501, that is, a correction target candidate is detected (S502). In this case, a region in which the upper end or the lower end of the region is aligned and the width does not change much is the candidate region. In addition, when assuming the fusion of captions, one of the conditions is that a figure region exists in the vicinity. In the example shown in FIG.

If a candidate area is detected as described above,
Next, step S503 is executed. That is, the character area is extracted again using the character elements constituting the candidate area. At this time, a forced division position (see FIG. 6) along the column is assumed, and the generation of a row and the generation of an area in the direction perpendicular to the reference row direction are performed outside the forced division position. FIG. 6 is an enlarged view of the relevant portion in FIG. Rows 602 and 603 are generated by using the element 601 outside the column with respect to the compulsory division position. Then, the matching degree in the generated row is calculated (S503).

Subsequently, if the matching degree is equal to or higher than a preset threshold value, this area is formalized, and a character element inside the column from the remaining forced division position is used to generate another character area. (S504). As the consistency of rows,
For example, the inclination of the regression line of the center of gravity of each character element or the variance of the distance between elements can be used. When a plurality of rows are generated outside the column as shown in FIG. 6, a value that is inversely proportional to the difference in the height of the rows (vertical direction in FIG. 6) may be used as the matching degree.

Therefore, by the processing operation as described above, an accurate region can be extracted even when the caption shown in FIG. 7 is close to the main part.

By the way, the document image processing operation in each of the embodiments described above is executed by the system shown in FIG. 1. In addition, as a document image processing method, a document image is recorded on a machine-readable recording medium as software. It may be executed on a computer.

[0029]

As described above, according to the document image processing method according to the present invention (claims 1, 2 and 3), the positions of the columns extracted in the column extraction process are referred to and excessively integrated. In order to evaluate and correct the extracted character area, avoid over-integration of the character area and extract it as the character area that should be extracted, even for a document image in which characters belonging to different character areas are close to each other. Can be.

According to a fourth aspect of the present invention, there is provided a computer-readable recording medium storing a program for causing a computer to execute the document image processing method according to any one of the first to third aspects. Claim 1
The operations described in any one of (3) to (3) can be realized by a computer.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a system to which a document image processing method according to an embodiment of the present invention is applied.

FIG. 2 is a flowchart illustrating a procedure of a document image processing method according to the embodiment of the present invention.

FIG. 3 is an explanatory diagram showing an example of column information extraction according to the embodiment of the present invention;

FIG. 4 is an example of a histogram of an area width created and used when extracting column information according to the embodiment of the present invention.

FIG. 5 is a flowchart illustrating a detailed processing procedure of an area correction process in FIG. 2;

FIG. 6 is an explanatory diagram showing an enlarged view of FIG. 7 of an example of forced re-region extraction and evaluation according to the embodiment of the present invention.

FIG. 7 is an explanatory diagram showing a case where the caption of the figure and the text are close to each other.

[Explanation of symbols]

 Reference Signs List 101 Image input unit 102 Image reduction unit 103 Basic element extraction unit 104 Character region extraction unit 105 Column detection unit 106 Character region correction unit 107 Control unit 108 Data storage unit 110 Data communication unit

Claims (4)

[Claims] 1. A document image processing method for extracting a character region from a document image and correcting excessive integration of the extracted character region when performing character recognition or layout information acquisition for the character region. An image inputting step, and reducing the document image input in the image inputting step,
Extract the circumscribed rectangle of the connected components of the black pixels constituting the character,
A basic element extracting step of extracting a basic element, a character for classifying the basic element into characters, tables, figures, and others, generating a line by integrating the character elements, and extracting a character area by integrating the line An area extraction step, a column extraction step of extracting column information from the character area extracted in the character area extraction step, and a character column excessively integrated with reference to the position of the column extracted in the column extraction step. A document image processing method, comprising: an area correcting step of correcting an area. 2. The area correcting step detects a character area in which excessive integration is expected from the position of the column extracted in the column extraction step, and determines the position of the character or character string element in the character area. 2. The document image processing method according to claim 1, wherein the text area is integrated again with reference to the text area, the degree of consistency of the line with respect to a preset reference division position is evaluated, and the text area is corrected. 3. The area correcting step, when integrating into a character area again, first generates a character area constituted by a direction perpendicular to a reference line direction located outside the reference division position, and thereafter, 3. The document image processing method according to claim 2, wherein a character area is evaluated. 4. A machine-readable recording medium storing a program for causing a computer to execute the document image processing method according to claim 1.