JPH06195507A - Character segmentation device - Google Patents

Abstract

PURPOSE:To correctly segment even image data, which includes a sideline touching a character, into even the character touching the sideline by the character segmentation part of a character recognition device. CONSTITUTION:The character segmentation part 3 segments the image data, read by an image read part 1, into character image data by characters. A circumscribed rectangle is found for each character image data and its longitudinal-lateral ratio is checked. When a longitudinal-lateral ratio is remarkably different, the histogram of black pixels in the character image is generated. When there is a black pixel area which projects nearby an end part in the character image, it is judged that there is a sideline. When it is judged that there is the sideline, image data other than the side line is segmented into characters again. Consequently, the character image having the sideline touching the character can correctly be cut.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention reads print, dot character and handwritten character patterns of newspapers, magazines, novels, etc. as image data from an image reading device such as a scanner and converts each character into code information such as JIS code. The present invention relates to a character slicing device used for a character recognizing device.

[0002]

2. Description of the Related Art Generally, a character recognition device has an input stage of optically scanning and reading a print character, a handwritten character, etc., and converting it into an electric signal, detecting character image data of the electric signal, cutting out, normalizing, A pre-processing stage such as noise removal,
It is roughly divided into a feature extraction step of extracting the characteristics of the character image data and an identification step of determining the character image data based on the characteristics by a method such as pattern matching.

In particular, the most important character pattern cutout is included in the preprocessing stage, and the character pattern cutout is performed by a character cutout device for separating a character pattern for one character from a character line from other character patterns. This character slicing device uses the character pitch and space information between characters to separate characters, and for handwritten characters, character slicing is performed using a character entry frame.

[0004]

In the conventional character slicing device, since one continuous region of black pixels is regarded as character image data for one character, underscores that contact a character as shown in FIG. 9A, for example. If there is a line, "A",
Characters "T" and "E" are regarded as continuous areas, and the cutout result is cut out as a single character "ATE" as shown in FIG. 9B, and a correct cutout result cannot be obtained. This occurs in both vertical writing and horizontal writing, and there is a problem in that it is not possible to accurately cut out characters to which a side line is attached one by one.

The present invention has been made in order to solve the above-mentioned problems, and it is possible to accurately cut out each character even if it is character image data in which a character to be cut out has a side line. The purpose is to provide a device.

[0006]

A character cutout device according to the present invention cuts out image data read from an image reading device into character image data for each character, and cuts out the character data as character image data. When the aspect ratios of the circumscribed rectangles of the two character images are significantly different, a histogram of the black pixels in the character image is taken, and in the histogram the presence or absence of a black pixel region protruding near the end of the character image It is determined whether or not a line is attached to the line, and the character is again cut out using only the image data other than the line based on the determination.

[0007]

According to the present invention, according to the above-described structure, a lateral line is included in a character image based on the ratio of the aspect ratio of the extracted character image and the presence or absence of a protruding black pixel area in the black pixel histogram of the character image. It is determined whether or not the character image data for each character is correctly cut out from the character image data other than the line.

[0008]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram of a character slicing device according to an embodiment of the present invention. In the figure, the character cutting device includes an image reading unit 1 for photoelectrically converting a recognition target document, a line cutting unit 2 for cutting image data of a character line unit from image data of the recognition target document, and a cut-out character line unit. A character cutout unit 3 that cuts out character image data in character units from image data, an underline determination unit 4 that determines whether there is an underline in the cut out character image data, and image data of an underlined portion in the character image. Backup unit 5 that backs up and stores
And an underline erasing unit 6 for erasing the image data of the backed up portion from the original image, a character re-clipping unit 7 for again performing character clipping for the image data with the underline erased, and a backup. The configuration includes an image restoration unit 8 that restores the image data of the underlined portion to the original image, and an end determination unit 9 that determines whether or not all the lines have been processed.

FIG. 2 is a detailed block diagram of the underline determination unit 4. In the figure, the underline determination unit 4
Is a circumscribing rectangle ratio detecting unit 10 that detects an aspect ratio of a circumscribing rectangle of one character image of the character image data cut out by the character slicing unit 3, and determines that the detected aspect ratio is equal to or larger than a predetermined value. The displayed character image is m (m is a positive integer)
A character image division unit 11 that divides the image into equal parts, a histogram calculation unit 12 that calculates a histogram of black pixels in the character image for each of m character images, and a lower part of the character image if the histogram is a horizontally written document. In the case of a vertically written document, the side-line determining unit 13 determines that an underline or a side line is attached to the character image if a black pixel area protruding to the right of the character image exists.

The operation of the character slicing device of this embodiment constructed as described above will be described with reference to FIGS. 3, 4, 5, 6, 7 and 8. The original image is assumed to be a horizontal writing document.

FIG. 6 is a flow chart of the character slicing device in one embodiment of the present invention.

In FIG. 6, the image reading section 1 optically reads out a document to be recognized and performs photoelectric conversion to obtain image data (21). The line cutout unit 2 cuts out the image data line by line from the image data (22). The cut-out line-by-line image data is output to the character cutout unit 3, and the character cutout unit 3 focuses on one of the plurality of cutout lines (23) and performs character cutout on the focused line (24). ).

Next, the underline determination unit 4 shifts to an operation for determining the presence or absence of an underline in the cut out plural character image (20). FIG. 7 shows the details of the operation flow chart for determining the presence or absence of underline. As a pre-process of this determination operation, the circumscribing rectangle ratio detection unit 10 focuses on one character image of the plurality of character images (25) and obtains the aspect ratio of the character image (26).

Next, the underline determination unit 4 checks whether the width of one character image is more than twice the height (27), and if it is less than twice, moves to the operation (32) described later.

When the underline determination unit 4 determines that the aspect ratio of one character image is 1 to 2 or more (27), it further checks whether or not there is an underline in the character image (28). .

FIG. 7 is a flow chart of the underline determination unit of the character slicing device in one embodiment of the present invention.

The specific determination of the presence or absence of underline will be described with reference to the flowchart shown in FIG.

FIG. 3 is a black pixel distribution diagram in the divided cells in the underline determination unit of the apparatus shown in FIG.

The character image division unit 11 is a circumscribed rectangle ratio detection unit 1.
A character image having an aspect ratio of 2 times or more detected at 0 is divided into m equal parts as shown in FIG. 3 (41). The histogram calculation unit 12 obtains the percentage of black pixels in each square made up of rows each having a number of dots corresponding to the number of pixels in the vertical direction of this character image and columns in the horizontal direction equally divided into m (42).
This result is shown in FIG. 3, and at the same time, a histogram based on FIG. 3 is prepared as shown in FIG. The side line determination unit 13 determines the number of dots (rows) in the created histogram.
Search for a row in which all 1 rows of 100% are black pixels (4
3) If there is a line with 100% black pixels, it is determined that there is an underline (46). The side-line determination unit 13 is 1
The length of the underline is determined by the width of the continuous divided grid area consisting of a row of 00% black pixels (44).

FIG. 5 is a black pixel histogram based on the black pixel distribution chart of the apparatus shown in FIG. How to determine the length of this underline will be described with reference to FIG. 5 in the figure
10 in row 5, column 1 to the right of row 0, column 100
There is 0. In contrast to 100 in 5 rows and 1 column, 100 exists in 6 rows and 3 columns in the lower right corner. In this way, 1 to the front, back, left, right, upper right, lower right, upper left, lower left of the target 100
If 00 is present, it is regarded as continuous. In FIG.
The width of the continuous underline is 4.

Further, the side-line judging unit 13 judges whether the continuous width of 100 is 90% or more of the entire width of the character image (4
5). In this determination, if it is 90% or more, it is determined that there is underline (46), and if it is 90% or less, it is determined that there is no underline (47).

When it is determined whether or not there is an underline, the process returns to the operation (30) and the operation (32) in FIG. 6 and the processing is restarted.

If the side-line determining unit 13 determines that there is no underline, the operation proceeds to an operation (32) for determining whether or not the character image of the line of interest has an unprocessed character image. If it is determined that the underline exists, the underline character image data from the focused character image data is backed up and stored in the backup unit 5, and then the underline erasing unit 6 erases the underline image data from the original image. (30) (31). This underline erase operation will be described with reference to FIG.

Here, FIG. 8 is a flowchart of the underline erasing unit of the character slicing device in one embodiment of the present invention.

In FIG. 8, the histogram calculation unit 12
The underline erasing unit 6 finds the row with the largest number of black pixels based on the histogram obtained in step (50), substitutes the number of the largest number of black pixel rows in row i (51), and subtracts 1 from row i ( 52). Next, it is checked whether or not the 100% portion exists in the i-th row (53), and if it is determined that the 100% portion does not exist, the operation shifts to the operation (56). Also, the black pixel length of row i is i + 1
It is checked whether the length is less than half the black pixel length of the row (54), and if the conditions are met, the operation moves to (56).

Next, it is checked whether the black pixel length of the i-th row is larger than the black pixel length of the i + 1-th row (55), and if the conditions are met, up is determined.
i + 1 is substituted for per (56), and u in the target character image
The images below the pper line are backed up and stored in the backup unit 5 (57).

Next, the underline erasing unit 6 erases the image below the upper line in the character image of interest (58).

When the image erasing operation is completed, the operation will be described by returning to FIG. 6 again. In the figure, it is checked whether or not there is an unprocessed character image in the line of interest (32), and if it exists, the process is executed again by focusing on one character image (2).
5).

Next, it is judged whether there is an underlined character image (33), and if there is not, a judgment is made as to whether there is an unprocessed line (36). When it is determined that the underlined character image exists, the character is cut out again for the line of interest (34).

Next, the image restoration unit 8 restores the image data backed up and stored in the backup unit 5 to the original image (35). The end determination unit 9 checks whether or not there is an unprocessed line in the restored original image (36). When it is determined that the character exists, the process returns to the character cutout unit 3 and proceeds to the character cutout operation (23). If it is determined that there is no unprocessed line, the character cutout operation ends.

In the above embodiment, the character cutting is performed again after the underline image is erased. In another embodiment, the position information of the underline image area is extracted and the information is extracted by the character cutting unit 3. By sending the data to the user, the correct character cutout result can be obtained even if the character cutout is performed again using only the image other than the underlined image area.

Further, in the above-described embodiment, the character re-cutout unit 7 performs the character cutout again after the underline image is erased. However, the character cutout unit 3 for executing the character cutout operation first also performs the character cutout operation again. May be performed in.

Further, although the underline determination unit 4 is configured to include the character image division unit 11, the circumscribed rectangle ratio detection unit 1
It is also possible to directly output the character image whose aspect ratio is 0 or more and the aspect ratio of which is equal to or greater than the predetermined value to the histogram calculation unit 12, and the histogram calculation unit 12 calculates the black pixel histogram of the character image.

Further, in the above embodiment, the aspect ratio of the circumscribed rectangle is set to 1: 2 as a condition for underline determination, but it can be set to any ratio according to the type of input character or the identification level.

[0036]

As described above, according to the present invention, based on the ratio of the aspect ratio of the cut out character image and the presence or absence of the protruding black pixel area in the black pixel histogram of the character image,
Since it is determined whether the character image includes a lateral line, the character image data for each character is properly cut out from the character image data other than the lateral line.

[Brief description of drawings]

FIG. 1 is a block diagram of a character clipping device according to an embodiment of the present invention.

FIG. 2 is a block diagram of an underline determination unit in the character clipping device of the device shown in FIG.

FIG. 3 is a black pixel distribution diagram in a divided grid in the underline determination unit of the apparatus shown in FIG.

FIG. 4 is a black pixel histogram based on a black pixel distribution chart of the apparatus shown in FIG.

5 is 100% based on the black pixel distribution chart of the device shown in FIG.
Black pixel distribution map

FIG. 6 is a flowchart showing a control procedure of the character cutting device according to the embodiment of the present invention.

FIG. 7 is a flowchart of an underline determination unit of the character cutting device according to the embodiment of the present invention.

FIG. 8 is a flowchart of an underline erasing unit of the character slicing device according to the embodiment of the present invention.

FIG. 9A is an original diagram of character data for explaining the problems of the conventional technology. FIG. 9B is a diagram showing a character cutout result mode for explaining the problems of the conventional technology.

[Explanation of symbols]

 1 image reading unit 2 line cutout unit 3 character cutout unit 4 underline determination unit 5 backup unit 6 underline deletion unit 7 character recutout unit 8 image restoration unit 9 end determination unit 10 circumscribed rectangle ratio detection unit 11 character image division unit 12 Histogram calculation unit 13 Side line determination unit

Claims (2)

[Claims] 1. An aspect ratio of a circumscribing rectangle of a character cutout unit that cuts out image data read from an image reading device into character image data for each character and one character image of the character image data cut out by the character cutout unit. A circumscribing rectangle ratio detecting section for detecting, a histogram calculating section for calculating a histogram of black pixels in a character image whose aspect ratios detected by the circumscribing rectangular ratio detecting section are remarkably different, and an edge of the character image in the histogram calculating section And a side line determination unit that determines whether or not a side line is attached in the character image based on the presence or absence of a black pixel region protruding in the vicinity of the area, and based on the determination of the side line determination unit A character slicing device characterized by slicing out character image data again. 2. A character image dividing unit that divides a character image, which is determined by the circumscribing rectangle ratio detecting unit to have an aspect ratio of a predetermined value or more, into a plurality of equal parts, and outputs the divided character image to a histogram calculating unit. The character slicing device according to claim 1, wherein