JPS60110085A - Character segmenting device of optical character reader - Google Patents

Abstract

PURPOSE:To segment one by one the characters within a character entry frame which is written in a non-drop-out color by searching the inside area of the character frame after detecting the upper left, upper right, lower left and lower right corners of said frame set in a grid form. CONSTITUTION:A character entry frame 2 written on a slip 1 is scanned by a photoelectric conversion element 3 together with the characters written within the frame 2. The analog signals thus obtained are converted into binary coded signals by an A/D converter 4, and data equivalent to one line on the slip 1 are stored to a line buffer memory 5. A rule detection part 7 scans a pattern stored in a line buffer 5 to detect a rule. A control part 6 performs the control so that the vertical center position of the frame 2 equivalent to one line is set at the center position of the Y direction of the buffer 5 by the format information. Then the part 7 calculates both right and left edge addresses x2 and x1 of right and left edge fields of the frame 2 fed to the buffer 5 by the position information Fs and Fe on the frame 2 which are previously given to the part 7 from outside. Thus rule detection reference points A-D are obtained. Then a character segmenting part 8 segments one by one the characters within the frame 2 which is written in a non-drop-out color.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a character cutting device, and more particularly to a character cutting device in an optical character reading device that optically reads characters written on a form.

Optical character reading equipment M (hereinafter referred to as OCR) generally uses optically undetectable (so-called drop-out colors) that eliminate the ability to read a single character on a form with a grid-like character frame. Since it is necessary to print the grid character frame in dropout color, the disadvantage is that it is necessary to create a form for OCR using a copier capable of multicolor printing. There is.

SUMMARY OF THE INVENTION Therefore, an object of the present invention is to provide a character cutting device in one oca that can cut out characters even from a form in which a character entry frame is printed in an optically sensitive color system (non-dropout color).

The OCR character cutting device according to the present invention includes a grid-like character entry frame and a line buffer memory that optically scans each character group written in the character entry frame in about one line year and stores the digitized data. The line buffer memory is searched using format information indicating the position of the character entry frame specified from the outside to detect the vertical and horizontal ruled lines of the character entry frame, and the position information of the detected vertical and horizontal ruled lines is used. It is characterized by having means for cutting out characters one by one in the character entry frame.

The present invention will be explained below with reference to the drawings.

FIG. 1 is a diagram showing an example of a book that can be read by the loading method of the present invention, where 11 is a form, 2 is a grid-like character entry frame written in non-dropout color, and Ll is a Distance from the top of the form to the center position of the character entry frame on the first line, P is the pitch between the character entry frame and the next line's character entry frame, FSll-It is the left edge of the leftmost field from the right edge of the form to the leftmost character entry frame , Fe is the distance from the right side of the form to the right edge of the rightmost field in the character entry frame, Fm
m indicates the distance from the right edge of the form to the left edge ruled line of the (n+1)th field of the mth line character entry frame. However, the value of n may be different for each row. .

FIG. 2 is a block diagram showing an embodiment of the present invention, in which the character entry frame 2 written on the form 1 and the characters written in the frame are scanned by a photoelectric and disguise element 3, and the result is obtained. convert the analog signal into a binary signal with an A-D converter 4,
One line of data on the form is stored in the line buffer memory 5. A single ruled line detection unit 7 scans the pattern stored in the line buffer 5 and detects a ruled line.

The method for detecting ruled line 1vlli1 will be explained with reference to FIG. 3. FIG. The control section 6 controls the character entry frame given in advance to the rt; Based on the position information Fs and Fe, the line detection unit 7 calculates the left end address X of the leftmost field and the right end address X of the rightmost field of the character entry frame input to the line buffer 5, and uses the ruled line detection criteria Points A, B
, C, and D. A, A, the upper left corner of the character entry frame, B point is the upper right corner of the character entry frame, 0 point is the lower left corner of the character entry frame, and D point is the lower right corner of the character entry frame. This is a reference point for detecting the intersection of a vertical ruled line and a horizontal ruled line, and the address of each point is determined by the following equation (1).

In equation (1), the subscript y is the X-direction address of the line buffer 5, the subscript X is the xjj-direction address of the line buffer 5, the height of the line buffer 5, P is the pitch of the character entry frame, and α and β are appropriate values. Let it be a constant.

To detect the intersection of the vertical and horizontal ruled lines at the upper left corner of the character entry frame, move U in the Y direction from point A determined using equation (1).

The number of black bits is counted while scanning a rectangular area Ra (hatched area indicated by Ita in FIG. 4(a)) of t in the X direction in the X direction and the X direction, and a histogram of each scan is created. (u*tFi is an appropriate constant). FIG. 4 (+) shows the histogram in the X direction. The address Ay' of the center value of the width Way of the portion where the number of black points exceeds θ in the histogram is registered as the X address of the horizontal ruled line at the upper left corner of the character entry frame.

FIG. 4(C) shows a histogram in the X direction. Figure 4 (
In C), let the address Ax' of the center value of the width Wax of the part where the number of black dots exceeds θ be the X address of the vertical ruled line at the upper left corner of the character entry frame, and the intersection point A / () of the vertical and horizontal ruled lines at the upper left corner.
, xl, Ay/> are registered.

In a similar manner, the intersection point B' (Bx) of the vertical and horizontal ruled lines at the upper right corner.

By') * Intersection of vertical and horizontal ruled lines in the lower left corner C'(Cx', Cy
').

The intersection D' (Dx t Dy) of the vertical and horizontal ruled lines at the lower right corner is registered.

A/, f31, C/ registered using the above method
, D/ are transferred to the character cutting section 8. This character cutting section 8 vertically searches the range of the character entry frame shown in FIG. 5 using the method described below, thereby saving the need for cutting out characters. In Fig. 5, the X coordinate of xmm is Ax'
-Frrm). The character cutting part 8 is the fifth
Regarding the X coordinate of the first field of the row buffer shown in the figure, a search is performed in the range of (Ax'+q) to (Xm do q) q: constant. Further, regarding the Y direction, by searching the range shown by the following equation (2), it is possible to prevent the character entry frame from invading the search area.

In equation (2), γ is a constant, and X is the X coordinate on the row buffer at the start of the search.

As for the X coordinate of the last field of the row buffer, the range from (Xmm+Q) to (Bx'-q) is searched. Also, in the Y direction, the same <(2
) By searching the range of the expression, it is possible to prevent the character entry frame from invading the search area.

For the X coordinate of the intermediate field excluding the first field and the last field, search the range from (Xml+q) to (Xml+t-Q).However, the value of 1=1 to (n-1) is searched. In addition, in the Y direction, by similarly searching the range of equation (2), it is possible to prevent the character entry frame from invading the search area.

The distance of the X coordinate from the Y-direction scan address where black is first detected as a result of the search in each field to the next inflection point from black to white 1i1 [t CWk (k is the block of the cut out point part) CWlc≧K ・・・・・・・・・・・・ If (3) is satisfied, it is assumed that characters exist in that range, and the image between them is transferred to the character recognition unit 9. .

For example, FIG. 6 shows a part of the search area in FIG. 5, and for the first character 1I211, the point where black is first detected is Xa, and the point where black changes to white. is Xb, and CW□=Xb-Xa. Next, for the noise point where black is detected, FiCW2=X(1-
X c, and the next character 1131 is CW,
:Xf-Xe. If K is set to a suitable value, then CW is ≥, CW2 (K, CW3 is ≥, and the character 1 is set.
The character image of 121.1311 is transferred to 9 and n2
--3'' is transferred to the character recognition unit 9.

As shown above, according to the present invention, it is possible to search the internal area of a character entry frame by detecting the upper right, upper right, lower left, and lower left of the character entry frame set in a grid pattern. It is possible to cut out characters one by one inside the character entry frame written in non-dropout color. Therefore, it is possible to use an OCR form printed by a copying machine or the like which cannot perform multicolor printing as the form.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing an OCR form that enables character extraction according to the present invention, and FIG. 2 is a block diagram of an embodiment of the present invention.
Figure 3 shows the upper left and upper right of the text entry frame. A diagram to explain how to detect the bottom left and bottom right respectively,
Figure 4 is a diagram showing an example of a histogram of the number of sunspots with 8 points to detect the upper left corner, Figure 5 is a diagram to explain the method of searching inside the blank area where a character entry frame has been detected, and Figure 6 The figure is a diagram for explaining a method of separating characters. 1 Explanation of the symbols of the sword part 1...Book W, 2...Character entry frame, 3.
...Photoelectric conversion unit, 5...Row buffer memory, 7...Rule iV! Detection section, 8...Character cutting section (b), 4 Δ)! (Att) (C) Figure 4 Figure 5 and Figure B

Claims (1)

[Claims] An optical method that stores the digital signal obtained by optically scanning a form line by line and cuts out characters one by one based on format information indicating the position of the character entry frame on the form. In the character cutting device of the character reading device,
A character entry frame set in a grid pattern, a means for storing a group of characters entered in this character entry frame into the memory in one line, and the positions of vertical and horizontal ruled lines of the character frame which have been memorized. A character cutting device comprising means for detecting information and cutting out characters within the character writing frame based on the detected position information.