JPH0628520A - Character recognition device - Google Patents

Abstract

PURPOSE:To provide a character recognition device capable of correctly extracting a character area within a ruled line. CONSTITUTION:A character rectangle decision part 10 classifying picture data into a character rectangle or into others, a table rectangle decision part 12 extracting a linear component in the rectangle as to data except for the character rectangle and judging whether it is a table or others except for the table from the linear component, and a ruled line area extraction part 13 detecting circumscribing rectangle black element density and whole black picture element density in a rectangle area as to circumscribing rectangle of the others except for the table and judging whether the area is the ruled line area or a graphic area from black picture element density and linear component information are provided. Rectangle information on the rectangle judged to be the ruled line area is exchanged with the extracted linear component.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention captures a document image by using an optical means such as a scanner, etc., for creating a database of printed documents and reusing the document, and characters, figures, tables, etc. are captured from the captured image data. Extract the area for each attribute such as ruled line,
The present invention relates to a character recognition device that performs recognition processing according to each attribute.

[0002]

2. Description of the Related Art A conventional method will be described below.

First, a circumscribing rectangle is detected from binary data read by a scanner, and is classified into a character rectangle and a non-character rectangle based on the size of the circumscribing rectangle and the black pixel density of the circumscribing rectangle. The character area is extracted by integrating.

For rectangles other than characters, if a minute rectangle is within a predetermined threshold value or if the black pixel density in the rectangle is greater than or equal to a predetermined threshold value, the rectangle The area becomes an image area.

With respect to the remaining rectangles that have not become characters / images, ruled line candidates within the rectangles are detected, and a region is extracted by determining whether they are tables or figures based on the line information.

When the extracted area overlaps another area, the areas are integrated into one area, and character recognition is performed on the character area. The image area is image-compressed, the table area is table-recognized, and the graphic area is vectorized.

[0007]

However, in the conventional method, in the example shown in FIG. 3, the area formed by a plurality of ruled lines as shown in FIG. Therefore, there is a problem that the internal characters are vectorized.

The present invention solves the above problems, and an object of the present invention is to provide a character recognition apparatus that correctly extracts a character area surrounded by ruled lines.

[0009]

In order to achieve the above object, the present invention provides a circumscribed rectangle black pixel density and a total black pixel density in the circumscribed rectangle (in the example of FIG. The rectangular black pixel density is the ratio of the number of black pixels of the ruled line in the area to the total number of pixels in the circumscribed rectangular area, and the total black pixel density in the rectangular area is the ruled line in the circumscribed rectangular area relative to the total number of pixels in the circumscribed rectangular area. And the ratio of the total number of black pixels). If the total black pixel density is equal to or greater than a predetermined multiple of the circumscribed rectangle black pixel density, and the line component extracted by the line component extraction unit of FIG. 2 is present in the figure candidate circumscribed rectangle, the line component is determined to be a ruled line and stored. The configuration is such that the figure candidate rectangle from which the ruled line is extracted is deleted.

[0010]

According to the present invention, since the ruled lines can be taken out one by one with respect to the region in which a plurality of ruled lines are combined, the character region surrounded by the ruled lines is also integrated into the ruled line region. It operates so as to be accurately extracted as a character area without being deleted.

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing the hardware configuration of an apparatus that executes area division in this embodiment. In FIG. 1, reference numeral 1 is a central processing unit (hereinafter, referred to as CP
2) and the image data input unit 7, the image data storage unit 8, the circumscribed rectangle detection unit 9, the character rectangle determination unit 10, the line component extraction unit 11, the table rectangle determination unit 12, and the ruled line region extraction unit shown in FIG. 13 and a recognition processing unit 14. 2 is a read only memory (hereinafter, R
OM). Reference numeral 3 is a random access memory (hereinafter referred to as RAM) in which image data read by the scanner 4 is stored. Reference numeral 5 is a keyboard for giving commands to the CPU 1 from the outside, and 6 is a CPU
1 is a display device that displays the recognition result recognized by 1.

The area division processing will be described below with reference to FIGS. 1 to 10. The image data input unit 7 and the image data storage unit 8 store the binary image data captured by the scanner 4 in the RAM 3 (s1 in the flowchart of FIG. 6). As shown in FIG. 9, the coordinates of the image data are such that the upper left corner is the origin, the horizontal coordinates are x, and the vertical coordinates are y.

The circumscribed rectangle detection unit 9 detects the circumscribed rectangle of black pixels and the number of black pixels connected in the vicinity of 8 from the stored image data (s2), and the circumscribed rectangle data, that is, the upper left coordinate. (X1, y1) and lower right coordinates (x2, y
2) and the number of black pixels are stored in the RAM 3. The circumscribed rectangle of the image data shown in FIG. 9 is shown in FIG. 10, and the coordinates (x1, y
1, x2, y2) = (3,2,7,8) and the number of black pixels is 1
5 is stored. As for the circumscribed rectangle of FIG. 3, ten circumscribed rectangles are detected as shown in FIG.

In the character rectangle determining unit 10, the length of the short side of the circumscribed rectangle is a predetermined threshold value Th1 (= 2).
5) or more, or if the ratio of black pixels in the area is less than or equal to the threshold Th2 (= 15), it is determined to be a rectangle other than a character, and the process is continued. The short side of the circumscribed rectangle is the threshold Th.
Is less than 1 and the circumscribed rectangle black pixel density is the threshold Th2.
If it exceeds, it is determined to be a character area, and the process proceeds to s9 (s3). The rectangle 1 in FIG. 4 is a rectangle other than a character and the process proceeds to s4, and the rectangles 2 to 10 are character candidates and the process proceeds to s9.

Here, the black pixel density is [black pixel density d1 =
Number of black pixels in circumscribed rectangle / (width of rectangle x height of rectangle) x 10
0%], the rectangle 1 in FIG. 4 has coordinates (x1, y1, x2, y2) = (5, 5, 34, 4).
4) and the number of black pixels is 98, the black pixel density d1 is
[98 / ((34-5 + 1) × (44-5 + 1)) × 1
00 = 8.17].

With respect to a rectangle remaining as a rectangle other than characters, a line component detection process for checking whether or not there is a line component in the rectangle is performed by the line component extraction unit 11 (s
4). The line component extraction method checks whether the length of the black pixel is equal to or greater than the threshold value Th3 (= 10) in each of the horizontal direction and the vertical direction. The result of extracting the line component of rectangle 1 in FIG. 4 is as shown in FIG. Based on the line components detected in this way, the table rectangle determination unit 12 next makes a table determination. The threshold value Th4 (= 4) in which the length of the detected horizontal line is the width of the rectangle
/ 5) times the threshold value is Th5 (= 3) or more, and the length of the vertical line is the threshold value Th4 times the height of the rectangle or more, the threshold value Th5 or more, and , The line crossing any of the above lines is the threshold value Th6 (= 2).
If so, the rectangle is determined to be a table (s5). A rectangle determined to be a table jumps to s9, and a rectangle not determined to be a table proceeds to ruled line determination processing. The rectangle 1 in FIG. 4 does not satisfy the conditions in the table because the line components are horizontal lines 2 and vertical lines 1 as in FIG. Therefore, the ruled line region extraction unit 13 performs ruled line determination described below on the rectangles that are not determined to be this table.

The circumscribed rectangular black pixel density detected in s3 and s6
It is determined whether the area is a ruled line area or not based on the relationship between the all-black pixel densities in the rectangle detected in step 1. The total black pixel density d2 is calculated by counting all black pixels in the rectangle, dividing the number by the area of the rectangle, and multiplying by 100. Rectangle 1 in FIG.
, The total number of black pixels is 239, so the total black pixel density d2
Is [239 / ((34-5 + 1) × (44-5 +)
1)) × 100 = 19.9] (s6). Is the total black pixel density d2 detected in s6 twice or more than the black pixel density d1 and the horizontal line detected by the line component extraction unit 11 is the threshold value Th7 (= 4/5) times the rectangular width? , Or
If there is a vertical line with a height Th7 times the height of the rectangle or more, the rectangle is determined to be a ruled line area rectangle (s7). Those that are not judged to be ruled line areas are judged to be graphic areas,
Jump to s9. For information that is determined to be a ruled line area, the rectangular information is deleted and the line information detected by the line component extraction unit 11 is added instead (s8). The rectangle 1 in FIG. 4 is determined to be a ruled line area candidate rectangle, the rectangle 1 is deleted, and lines 1, 2, and 3 are added as shown in FIG.

As a result of the above processing, the image data of FIG.
5 in total, which are the line and the ruled line, line 1, line 2 and line 3 in total 12
An area is extracted from each piece of rectangular information. The area result thus obtained is shown in FIG.

Next, it is checked whether the extracted areas do not overlap with other areas, and if they overlap, the areas are integrated (s9).

The recognition processing section 14 performs character recognition processing on the respective areas obtained as described above, after performing character cutting processing in the case of character areas. In the case of the graphic area, the graphic is vectorized, and in the case of the table area, the structure of the table is recognized, and the character recognition processing is performed on each cell.

As described above, according to this embodiment, the central processing unit for performing the area extracting process, the ROM storing the area extracting program, and the R storing the read image data.
The central processing unit includes an AM and a display device that displays a result recognized by the central processing unit. The central processing unit includes an image data input unit, an image data storage unit, a circumscribed rectangle detection unit, a character rectangle determination unit, a line component extraction unit, and a table. Includes a rectangle determination unit, ruled line area extraction unit, and recognition processing unit, classifies image data into character rectangles or other types, determines whether a table other than a character rectangle is a table or not, and determines a rectangle other than a table. It is determined whether a ruled line area or a graphic area is determined, and for the rectangle determined as a ruled line area, the rectangular information is converted into line components, and the ruled line areas are not extracted as one graphic area as in the conventional case, but one by one. Since it is extracted as the ruled line of, the character area surrounded by the ruled line can be correctly extracted.

In this embodiment, each threshold value is Th
1 = 25, Th2 = 15, Th3 = 10, Th4 = 4 /
5, Th5 = 3, Th6 = 2, Th7 = 4/5, but the values are not limited to these values.

[0023]

As is apparent from the above-described embodiments, according to the present invention, the ruled line area in which a plurality of lines are combined can be extracted as one ruled line without being extracted as one figure area. The character area surrounded by is not integrated into the graphic area, and can be accurately extracted as a character area, and a character recognition device capable of highly accurate recognition can be provided.

[Brief description of drawings]

FIG. 1 is a block diagram showing a hardware configuration of an apparatus that executes area division according to an embodiment of the present invention.

FIG. 2 is a block diagram showing a functional configuration of the device.

FIG. 3 is a schematic diagram showing an example of image data in this embodiment.

FIG. 4 is a schematic diagram of a circumscribed rectangle detection result of FIG.

5 is a schematic diagram of the line component extraction result of FIG.

FIG. 6 is a flowchart showing an image area dividing process according to the present embodiment.

FIG. 7 is an explanatory diagram of coordinates of image data.

FIG. 8 is a schematic diagram showing a region extraction result in this embodiment.

FIG. 9 is a schematic diagram showing an example of image data.

10 is a circumscribed rectangle diagram of the image data shown in FIG.

[Explanation of symbols]

 1 CPU 2 ROM 3 RAM 4 Scanner 6 Display Device 7 Image Data Input Unit 8 Image Data Storage Unit 9 Enclosing Rectangle Detection Unit 10 Character Rectangle Determination Unit 11 Line Component Extraction Unit 12 Table Rectangle Determination Unit 13 Ruled Line Area Extraction Unit

Claims (1)

[Claims] 1. A scanner for reading characters and generating image data, a RAM for storing the read image data, a ROM for storing a character area extraction program, a central processing unit for performing character area extraction processing, and the central processing. An image data input and storage unit for storing image data read by the scanner in the RAM, and a display unit for displaying the result recognized by the device, and the image data input and storage unit stored in the RAM. The circumscribing rectangle detecting unit that detects the circumscribing rectangle and the black pixel of the black pixel in the image data, and stores the coordinates of the circumscribing rectangle and the number of black pixels in the RAM, and compares the size of the circumscribing rectangle with a reference value to From the result of comparing the black pixel density, which is the ratio of the number of black pixels in the circumscribed rectangle to the total number of pixels in the rectangle, with the density reference value, the circumscribed rectangle is a character rectangle or a non-character A character rectangle determining unit determines the form,
A line component extraction unit for extracting a line component from the non-character rectangle determined by the character rectangle determination unit, and a horizontal line segment length and a vertical line segment length of the line component extracted by the line component extraction unit, respectively. A table rectangle determination unit that determines whether the non-character rectangle is a table from the result of comparison with a length reference value, and a rectangle determined to be not a table by the table rectangle determination unit with respect to the total number of pixels in the rectangle The total black pixel density, which is the ratio of the total number of black pixels in the rectangle, was compared with a total density reference value, and the horizontal and vertical line segment lengths extracted by the line component extraction unit were compared with respective length reference values. The central processing unit deletes the rectangle determined to be the ruled line region rectangle by the ruled line region extraction unit, and the line component extraction unit. Extracted Swapping the line components, and extracts the character region surrounded by ruled lines, configured character recognition apparatus as character recognition processing is performed.