JPH0387793A - Automatically character pattern shaping method - Google Patents

Abstract

PURPOSE:To automatically generate a sharp, high-quality character pattern that never deteriorates the shape of an adjacent one from an alphabetic character by adopting such a constitution that the noise of an adjacent edge part is removed when the linear noise of a character pattern is removed. CONSTITUTION:The numbers w1 and w2 of self picture elements in two noise lines adjacent to the horizontal line of a dot pattern which is generated by a scanner from the alphabetic character formed by a line inserting means are obtained, and if the rate w1 / w2 is less than a prescribed threshold, the recessed noise in the noise line is buried with a black picture element. On the other hand, if the ratio w1 / w2 exceeds the prescribed threshold, a line deleting means obtains an adjacent approximate straight line along the contour line with the following black picture element in the noise line adjacent to the horizontal line as a starting point, from a trace starting point and an already traced point, instead of removing the recessed noise, deletes black dots in a part protruding from an intersection C between the edge approximate straight line and the noise line, and removes the adjacent noise. Thus, a sharp, high-quality character pattern that never deteriorates an adjacent one can be automatically generated from the alphabetic character.

Description

[Detailed Description of the Invention] [Summary] Regarding an automatic character pattern shaping method that removes noise from the scale portions of letters, noise is removed from the edge portions of the scales when straight line noise is removed, and the shape of the scale portions is not distorted and is of high quality. The purpose of this paper is to automatically generate beautiful character patterns from the character capitals, and calculate the number of white pixels W, , W= in two noise lines adjacent to the horizontal line (or vertical line) of the dot pattern generated from the character capital using a scanner. Find these ratios w, /
A line insertion means for filling the concave noise in the first noise row with black pixels when wz is smaller than a predetermined threshold, and a line insertion means for removing the concave noise when the ratio W+/Wz obtained by the line insertion means is larger than a predetermined threshold. horizontal line (or vertical line) without doing
Trace along the contour line starting from the black pixel in the next row of the noise row adjacent to , find the edge approximation line of the scale based on the start point and this traced point, and connect this edge approximation line with the noise line. and a line deletion means for deleting black dots in the portion protruding from the intersection point C, and the line deletion means is designed to remove noise at the end portion.

[Industrial application field]

The present invention relates to an automatic character Bakun formatting method that removes noise from the end portions of characters. word processor,
In electronic typesetting systems, characters expressed in dot patterns are used for printing. In the case of Japanese, the number of characters used for these is enormous, and it is required to create them efficiently.

[Prior art and problems to be solved by the invention] Conventionally, handwritten characters are read in with a scanner, developed into a dot pattern, background noise removed, straight line noise removed, vertical and horizontal line shaping performed, and vectorized. I was trying to generate a character pattern.

However, since the scales were not treated appropriately, the shape of the end portion after shaping was distorted, resulting in a decrease in quality.

An object of the present invention is to remove noise from the edge portions of the scales when removing straight line noise, and to automatically generate beautiful character patterns of high quality from the capital letters without distorting the shape of the edge portions.

[Means to solve problems]

Means for solving the problem will be explained with reference to FIG.

In FIG. 1, the line insertion means 9 extracts the number of white pixels W, , w of two noise lines adjacent to the horizontal line (or vertical line) of a dot pattern generated using a scanner or the like.
2 are calculated, and when these ratios w1/w2 are smaller than a predetermined threshold, the concave noise in the first noise row is filled with black pixels.

The ratio W obtained by the line deletion means 10 and the line insertion means 9
When I/W2 is larger than a predetermined threshold, trace along the contour line starting from the black pixel in the next row of the noise row adjacent to the horizontal line (or vertical line) without performing concave noise removal, The edge approximation line of the scale is calculated using the starting point and this traced point, and the intersection point C of this edge approximation line and the noise line is calculated.
This removes the black dots that protrude from the image.

(Function) As shown in FIG. 1, the present invention provides the number of white pixels in two noise lines adjacent to a horizontal line (or vertical line) of a dot pattern generated by the line insertion means 9 using a scanner or the like. Find W 1 and W 1 respectively, and calculate their ratio Wl/
When Wl is smaller than a predetermined threshold, the concave noise in the first noise row is filled with black pixels, while the obtained ratio w, /w
2 is larger than a predetermined threshold, the line deletion means 10 traces along the contour line starting from the black pixel in the next row of the noise row adjacent to the horizontal line (or vertical line) without performing concave noise removal. , find the edge approximation line of the scale based on the starting point and this traced point, delete the black dots in the part that protrudes from the intersection C of this edge approximation line and the noise line, and remove the noise in the edge part. There is.

Therefore, by removing noise from the edge portions of scales at the same time as straight line noise removal, it is possible to automatically generate beautiful character patterns of high quality from the capitals without disturbing the shape of the edge portions.

〔Example〕

Next, the configuration and operation of one embodiment of the present invention will be explained in detail using FIGS. 1 to 3.

In FIG. 1, the letter l is a letter representing a character, and is neatly drawn on paper.

The graphic human input device t2 manually inputs a binary image generated by scanning the letter 1 using a scanner or the like into a computer or the like.

The input character pattern 3 is an image of the character 2 (1!) read by the graphic input device 2.

The background noise removing means 21 removes noise from the input cover turn 3, and is composed of 4 to 6.

Ambient noise removal means 4 is for removing noise generated around the character frames of letters and the like.

The isolated noise removal means 5 removes isolated noise that does not touch the character pattern.

The contour noise removing means 6 removes small noises of several dots that touch the character pattern.

The linear noise removing means 22 further removes linear noise from the character pattern from which the background noise has been removed by the background noise removing means 21, and is comprised of squares to 9.

The black run extraction means 7 extracts black runs in the horizontal and vertical directions from the character pattern.

The relationship extraction means 8 examines the mutual relationship of the black runs extracted by the black run extraction means 7, and extracts portions that appear to be horizontal lines or vertical lines from the characters.

The line insertion means 9 removes concave noise adjacent to the horizontal and vertical lines of the character extracted by the relation extraction means 8, and removes the white pixels of two noise lines adjacent to the horizontal line (or vertical &'jl). The number of W, ,w.

are calculated, and when these ratios W 1 /' W z are smaller than a predetermined threshold, the concave noise in the it noise row is filled with black pixels (described later with reference to FIG. 2).

The scale/noise removing means 23 is composed of a line removing means IO, and the ratio W I/W obtained by the straight line noise removing means 22 is
Horizontal line (or vertical line) corresponding to the chain part without removing concave noise when the noise is larger than a predetermined threshold.
Starting from the IN element of the next row of the noise row adjacent to This is to delete the black dots that protrude from the intersection C between the approximate straight line and the noise line (1! will be explained using FIG. 3).

The vertical/horizontal line shaping means 24 recognizes the vertical lines and horizontal lines of the character pattern from which noise has been removed by the scale noise removing means 23, etc., and aligns the line widths, and is adapted from 11 to 14.

The vectorizing means 11 vectorizes the outline of a character pattern.

The vector shaping means 12 removes noise existing in the vectorized contour vector.

The line width control means 13 recognizes vertical lines from contour vectors and controls the line width.

The full-dot converting means 14 restores the contour line of the character pattern from the contour vector and fills the inside thereof to generate a shaped character pattern.

The output character pattern 15 is a beautiful character pattern shaped by the vertical/horizontal line shaping means 24.

Next, the operation of the line insertion means 9 will be explained in detail using FIG. This is a noise line (Y, -1) and a noise line (Y□ This is to remove the concave noise of ゛+1〉.It will be explained below.

ST! ! P1 → Scan in the -Y direction to detect a black run that serves as a boundary above the horizontal line, and set the Y coordinate of that row to vl. Also, set the X coordinates of the left and right ends of the black run to x6, 3x
Set it to 2.

5TEP 2 X in (Yl-1) row and (Y,-2) row
, -X, and calculate the number of white pixels between W, , and W, respectively.
Let it be wt.

5TEP 3 (11w1/Ilt is greater than or equal to a certain threshold value, processing by the line deletion means 10 is performed.

(If 21w+/-o is below a certain threshold, cY14>
Line ×1~X! Fill in the gaps with 1 and remove concave noise (for example, fill in the dots indicated by vertical arrows as shown in "Result" in Figure 2).

5TEP 4 + Scan in the Y direction to detect a black run that is the lower boundary of the horizontal line, and set the Y coordinate of that row to Y. Also, let the X coordinates of the left and right ends of the black run be X8,×.

5TEP 5 X in (Y*+1) and (Y1+2) rows
Count the number of white pixels between s and Xe, and calculate W
Let j, w.

5TEP 6 (1) If Ws/wa is equal to or greater than a certain threshold value, processing by the line deletion means 10 is performed.

(If 21w s / w a is below a certain threshold, (
The space between X and x9 in the Yx+1) row is filled with 1 to remove concave noise.

ST'EP 7 + Scan in the X direction to detect a black run that is the left boundary of the vertical line, and set the X coordinate of that row to x1. Also, set the X coordinates of the left and right ends of the black run to YesY,
shall be.

5TEP 8 Y in rows (XI-1) and (X,-2)
, 〜Yt, and calculate the number of white pixels between W and Yt, respectively.
, vH.

5TEP 9 (1,,) If w1/ws is greater than or equal to a certain threshold, processing is performed by the line deletion means 10.

(2) If wt/wt is below a certain threshold, (L-1)
The spaces between Y and Yi in the rows are filled with 1 to remove concave noise.

5TEP 10 Scan in the +X direction to detect a black run that is the right boundary of the vertical line, and set the X coordinate of that row to XI. Also, let the X coordinates of the left and right ends of the black run be Yl, Y.

5TEP 11 Y in rows (XI-1) and (xz-2)
The number of white pixels between 's and Y* is counted and set as W= and W, respectively.

ST[! P 12 (1) If wi/wa is greater than or equal to a certain threshold, the line deletion means 10 performs processing.

12) If w x / w 4 is less than a certain threshold, the space between Y and Y4 in row (L-1) is filled with 1 to remove concave noise.

Through the above processing, for example, the character pattern shown in the upper part of Figure 2 is filled with dots indicated by vertical arrows in the character pattern shown in the lower part, concave noise adjacent to horizontal lines (or vertical lines) is removed, or w+/wx, wa/wa
If it is below a certain threshold, the next line deletion means 10 is executed to smooth the scale portion and generate a beautiful character pattern.

The operation of the line deletion means 10 will be explained in detail using FIG. When the shape around the scales is W+/Wt, W, /W4, which are described above, are greater than a predetermined threshold value, it is recognized as a scale part, and the following noise removal is performed on the chain part.

5TEP l Track black pixels adjacent to the horizontal line until their direction is reversed 0 For example, in Figure 3 ta+,
Draw a black dot on the outline starting from the starting point ■ of the next line of the noise line l adjacent to the horizontal line of the character pattern, ■, ■, ■, ■
, ■ and the tracking direction is reversed at ■, so the process ends.

5TEP 2 Find a straight line that approximates the edge of the scales based on the traced starting and ending points. This edge approximation straight line is created by calculating the distance between the straight line connecting the starting point ■ and the ending point ■, and each point ■ between them, and finding the ■ with the maximum distance (
Since this maximum distance is larger than a predetermined threshold value (for example, the distance of one pixel), the distance between the straight line connecting the starting point ■ and the obtained ■ is similarly calculated. is repeatedly performed until the distance becomes equal to or less than a predetermined threshold.

Here, a straight line connecting the starting point (■) and the point (2) is found, and this is used as the approximate straight line for the work Y (third opening)).

5TEP 3 Noise line 1 adjacent to the horizon and ST[
! The intersection point C with the edge approximate straight line extracted in step P2 is determined (Fig. 3 (C)).

5TEP 4 Remove black pixels outside the intersection C as noise. As a result, the black dot indicated by the arrow +dl in FIG. 3 is removed as noise, and the chain portion is smoothed by the edge approximation iii line.

〔Effect of the invention〕

As explained above, according to the present invention, since a configuration is adopted in which noise is removed from the edge portions of scales when straight line noise is removed from character patterns, high-quality and beautiful character patterns can be created without distorting the shape of the chain portions. It can be automatically generated from the alphabet.

[Brief explanation of drawings]

1 shows a composition diagram of one embodiment of the present invention, FIG. 2 shows a diagram for explaining line insertion, and FIG. 3 shows a diagram for explaining line deletion. In the figure, 9 is a line insertion means, IO is a line deletion means,
21 is a background noise removal means, 22 is a linear noise removal means, 2
3 represents a V4 noise removal means, and 24 represents a vertical/horizontal line shaping means.

Claims (1)

[Claims] In a character pattern automatic formatting method that removes noise from the scales of a character, white pixels in two noisy lines adjacent to a horizontal line (or vertical line) of a dot pattern generated from a character using a scanner etc. Find the numbers w_1 and w_2, respectively, and calculate the ratio w
a line insertion means (9) for filling the concave noise in the first noise row with black pixels when _1/w_2 is smaller than a predetermined threshold; and a ratio w_1/w_2 obtained by this line insertion means (9).
is larger than a predetermined threshold, without performing concave noise removal, trace along the contour line starting from the black pixel in the next row of the noise row adjacent to the horizontal line (or vertical line), and Find the approximate straight line of the scale edge based on the points,
A line deletion means (10) is provided for deleting black dots in a portion protruding from the intersection point C between the edge approximation straight line and the noise line, and the line deletion means (20) is configured to remove noise from scale portions. An automatic character pattern formatting method featuring the following.