JPS5932077A - Character segmenting device - Google Patents

Abstract

PURPOSE:To segment accurately characters, by segmenting a read pattern for each row to store them and counting the number of vertical black picture elements and the number of groups of continuous black picture elements to obtain the difference of count values as well as the changing state of said difference. CONSTITUTION:A pattern read out of a slip, etc. is segmented 1 for each row and supplied to a character segmenting device 2 to be stored in a picture memory 11. Then the number of black picture elements is counted 12 for each vertical column, and at the same time the number of groups of continuous black picture elements (black run) is counted 13. A shadow block is detected 14 out of the output of the counter 12, and at the same time a sudden change position is detected 15 for the count value. Then a region changing position is detected 16 from the output of the counter 13. The segmenting position is decided 17 from those detecting positions. The memory 11 is segmented at a segmenting part 18 and fed to a picture memory 3. For instance, the black run is constant for the 11th and 12th columns shown in the figure, and the black picture element has a width approximately equal to the width of a line to be defined as a segmenting position candidate. Thus it is possible to segment characters even though no space exists between characters.

Description

DETAILED DESCRIPTION OF THE INVENTION [Technical Field] The present invention relates to a character cutting device that cuts out a reading pattern of characters read from a form or the like, character by character.

[Prior art]

In a conventional character cutting device, a cutting method as shown in FIG. 1 is used.

■ A method of detecting the presence or absence of black pixels on each scanning line and cutting out all parts of consecutive black pixels as one character.

■ A method of counting the number of black pixels for each scanning line, determining the projected density distribution, and cutting out characters based on that distribution.

■ For characters with a constant character pitch, a method of cutting out characters using that character pitch.

However, as shown in Figure 2, when there is no space between characters, such as connected characters, the above method cannot be used, and when the character pitch is variable, the desired character cannot be cut out using the above method. The problem was that I couldn't do it.

〔the purpose〕

The present invention has been made in consideration of the above-mentioned circumstances, and an object of the present invention is to provide a character cutting device that can cut out desired characters even when the character pitch changes and the spaces between characters are wide. shall be.

〔Example〕

An embodiment of the present invention is shown in FIG. Characters written on a form or the like are read by a reading section (not shown), and one line of the reading pattern is cut out by a line cutting section l. Line Cutting Redundant reading patterns are cut out for each character stored in the image memory 11 of the character cutting device 2. The black pixel counter 12 counts the number of black pixels for each vertical scanning line of the reading pattern. The black run counter 3 counts the number of consecutive black pixel groups (referred to as "black runs") for each scanning line. The black pixel counter 12 and the black run counter 13 may be configured as one counter. Black pixel counter 12t black I for each scanned line! Using the il prime number, the projection block detection section 14 detects a "projection block," and the sudden change position detection section 15 detects a "sudden change position."
is detected. Further, the area change position detecting section 16 detects the "area change position" using the number of black runs for each scanning line counted by the black run counter 13. The "projection block", "rapid change position", and "region change position" will be described later. Using the detection results of these detection units 14, 15, and 16, the extraction determination unit determines the extraction position (details will be described later).
In step 8, the image is cut out at the determined cutting position. The extracted reading pattern for each character is stored in the character memory 3 and subjected to processing such as character recognition.

Next, the present embodiment will be described in more detail with reference to FIG. 4 and FIG. 5.

(1) The black pixel counter 12 counts the number of black pixels for each scanning line.

(2) The black run counter 13 counts the number of black runs for each scanning line.

(3) Based on the number of black pixels counted in (1), a range of continuous scanning lines in which the number of black pixels is not zero, that is, a "projection block" is detected by the projection block detection unit (4).

(4) The position where the number of black pixels counted in (1) changes by more than twice the character line width (“7” in this example) is defined as a “rapid change position” and is detected by the rapid change position detection unit 15. To detect.

(5) The area change position detection unit 16 sets the position where the number of black runs counted in (7) changes as the “area change position”.
Detected by

(6) If the area change position is not within a projection block, cut out the projection block as one character.

(7) The projection block obtained according to (3) is
In this embodiment, if it is smaller than r18J), it is cut out as one character.

(8) If the conditions (6) and (7) are not satisfied, perform the following processing in a "processing range" excluding the left and right quarters of the projection block.

(i) In an area where the number of black runs counted in (2) is 1, if there is a position where there is a sudden change in the density, divide the area at that position, and divide the area into the area where the number of black pixels counted in (1) is smaller. If there is a position where the number of black pixels is equal to or less than the line width (“3” in this embodiment), it is determined as a candidate for the cutting position. A in FIG. 4 is a candidate for the cutting position.

(b) If there is no sudden change position, in all areas where the number of black runs is 1, if there is a position where the number of black pixels is equal to or less than the line width ("3" in this example), it is considered as a candidate for the cropping position. .

(ii) In an area where the number of black runs counted in (2) is 2, both ends of this area are abrupt change positions υ, the number of pixels on both sides is large, and the number of pixels in this area is the line width. About twice as much (
In this embodiment, if there is a position equal to or less than "5", it is considered a cutting position candidate. B and C in FIG. 5 are cutout position candidates.

(fit) Cutting position candidates that are five or more away from the center O of the processing range are rejected.

(i■) When there are two or more candidate cutting positions, a position close to the center O of the processing range is set as the cutting position. In FIG. 5, of the cropping position candidates B and C, C closest to the center O is the cropping position.

(9) In (7), if the interval between two adjacent reading patterns cut out as one character is n or less and the combined width of the two reading patterns is m or less, 1
Cut out as text. In this example, n=51m=18.

As a result of the above processing, characters are cut out at position A in the reading pattern shown in Fig. 4 and at position C in the reading pattern shown in Fig. 5, resulting in a rounded lowercase English letter r L.
, 8 J p r t rj can also be correctly extracted.

Note that the numerical values for each judgment condition are determined empirically based on the characters to be cut out, the resolution of the sensor, etc.
It is not limited to the numerical values mentioned above.

〔effect〕

As described above, according to the present invention, even when the character pitch changes and there is no space between characters, desired character extraction can be performed.

It is particularly effective for printed characters with variable character pitch, and can be used in OCR, document editing devices, etc.

[Brief explanation of drawings]

Fig. 1 is a pattern diagram showing the character cutting method of a conventional character cutting device, and Fig. 2 (a), e, (b), (C), and (d) are pattern diagrams showing the character cutting method of a conventional character cutting device. FIG. 3 is a block diagram of a character cutting device according to an embodiment of the present invention, and FIGS. 4 and 5 are explanatory diagrams showing the character cutting operation of the same device. 1...Line cutting unit, 2...Character cutting device, 3...
・Character memory, 11... Image memory, 12... Black pixel counter, 13... Black run counter, 14... Projection block detection section, 15... Rapid change position detection section, 1
6... Area change position detection section, 17... Cutting out determination section, 18... Cutting out section. Applicant's agent Kiyoshi Inomata Figure 2 (a) (b) (c)
(d) Figure 3/2 Figure 4 (1) Number of black pixels 24 +52020t94211128
878891+ 13112 (2) Black number 12322
221223333333221 (3) Projection scolding□ (4) Sudden;
I (5) 9 pages! Shihen Ill Ji (8)
(i) Cutout 1Ω replacement A Fig. 5 (1) Number of black pixels 134162021206664481
616157677754 (2) Number of black runs 111111
12222231112221111 (3) Projection book □

Claims (1)

[Scope of Claims] A character cutting device that cuts out a reading pattern of characters read from a book board or the like into individual characters, comprising: a counter that counts the number of black pixels and the number of black runs for each scanning line in the reading pattern; , a projection block detection unit that detects a projection block based on the presence or absence of the number of black pixels in each scanning line counted by the counter; and a scanning unit in which the number of black pixels in each scanning line counted by the counter changes by a predetermined number or more. an abrupt change position detection section that detects a sudden change position of a line; an area change position detection section that detects an area change position of a scanning line where the number of black runs for each scanning line counted by the counter changes; and the projection Judgment conditions are set in advance based on the projection block detected by the block detection section, the sudden change position detected by the sudden position detection section, and the area change position detected by the area change position detection section. What is claimed is: 1. A character cutting device, comprising: a cut-out determination section for determining a cut-out position; and a cut-out determination section, which cuts out the reading pattern at the cut-out position determined by the cut-out determination section.