JPH0554139A - Contour tracking system for binary image - Google Patents

Abstract

PURPOSE:To unnecessitate a large capacity memory and to shorten processing time without retrieving a tracking direction each time of longitudinally and laterally enlarging a picture element double by tracking a contour picture element only by fitting the previously defined tracking direction to an inputted image. CONSTITUTION:The number of a central picture element is set as '9' and numbers from '1' to '8' are respectively attached to eight peripheral picture elements. When each picture element is black, a flag is turned to '1', when it is white, the flag is turned to '0' and each number attached to the picture element is expressed with the binary number of 9 bits as the digit of each bit. A number turning this number to a decimal number is defined as the pattern number of each pattern. Further, the flag of the picture element defining the tracking direction among the eight peripheral picture elements is turned to '1', and the respective picture element numbers are expressed with the binary number and hexadecimal number of 8 bits. A ROM or the like stores a table showing the tracking direction corresponding to these pattern numbers. Thus, since the tracking direction is defined for each pattern, the contour tracking of the image can be accelerated.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a contour tracing method for a binary image in which contours of an image (character pattern) formed as a set of pixels (dot pattern) are extracted and vectorized automatically. is there.

[0002]

2. Description of the Related Art Conventionally, a character read by a scanner or the like is converted into outline data for use, or a ROM is used.
In order to enlarge / reduce and use a character pattern formed as a dot pattern stored in a storage device such as the above, the contour of the character pattern is vectorized. Conventionally, various types of binary image contour tracking methods for automatically vectorizing have been considered. For example, JP-A-62-1
No. 31382 and Japanese Unexamined Patent Publication No. 64-91176, the input binary image is magnified twice in the vertical and horizontal directions and then the pixel is determined to be black (1) or white (0) to trace the contour. To search for the starting point for further movement, and further, in the clockwise, downward, leftward, rightward, and diagonally eight pixels in contact with the starting point pixel, and in the direction in which the pixel changes from 0 to 1, the boundary pixel forming the contour is detected. A method of tracking the center of the is disclosed.

[0003]

In the conventional contour tracking method for binary images, when the contour is traced after the binary image is enlarged vertically and horizontally twice, the image data amount is quadrupled. Had to be processed, which required a large amount of memory and processing time. And when tracking the border pixels,
When there are a plurality of boundary pixels in the tracking direction, it may be difficult to decide in which direction the tracking should be performed. In addition, when searching for a starting point or tracking a boundary pixel, since the direction in which the pixel changes from 0 to 1 is checked every time, there is a problem that the processing amount increases and it takes time. Therefore, an object of the present invention is to provide a contour tracking method for a binary image in which the amount of data is reduced and the processing time is shortened.

[0004]

As means for achieving the above object, in a contour tracing method of a binary image, which extracts some of boundary pixels which are contours of the binary image and vectorizes them, A contour tracking method for a binary image, wherein a tracking code corresponding to a pixel is attached to a pixel of the binary image, and vectorization is performed using the boundary pixels obtained by tracking the tracking code. It is the one we are trying to provide.

[0005]

According to the present invention, in a binary image, a flag for the tracking direction is defined by the presence or absence of 8 pixels surrounding a pixel (black and white discrimination), and this flag is added to all pixels to determine whether or not it is a boundary pixel. It discriminates and tracks along the tracking direction flag attached to this boundary pixel.

[0006]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the contour tracking system for binary images of the present invention will be described with reference to the drawings. First, the tracking direction from the black pixel to the adjacent black pixel is defined by discriminating each of the eight white / black pixels around the black pixel. At that time, the tracking direction is set to 2 directions (4 directions at the common point described later, 0 direction when the pixel does not become a boundary pixel), the inside of the block of black pixels is not tracked, and only one point is touched. The black pixel in the diagonal direction is not tracked, and the tracking direction to the adjacent black pixel is determined. That is, when there is an image composed of black pixels as shown in FIG. 1A, the tracking direction of the boundary pixel a is the direction a1 in FIG. , a2, and do not trace in the direction of a3 to a5, which is the direction toward the inside of the block of black pixels, and the direction to the white pixels. Similarly, the tracking direction of the boundary pixel b in FIG. 1A is defined as the directions b1 and b2 in FIG. 1C showing the boundary pixel b and the eight surrounding pixels, and only one point is defined. Do not trace the diagonal black pixels (direction b3) that are in contact with each other. In this way, for example, FIG.
As shown in (F), two tracking directions are defined so that the boundary pixels are tracked for all the patterns of the white / black pixels of the eight pixels surrounding the pixel. At this time,
The patterns shown in (A) and (C) rotated 90 ° counterclockwise and the patterns shown in (F) and (D) of FIG. 2 match, so that the tracking directions also match. The target patterns are defined to have the same tracking direction when they rotate and become the same pattern.

Then, in order to store the defined tracking direction for each pattern, the following pattern numbers are given to all patterns. That is, as shown in FIG.
The number of the central pixel is set to 9, and the surrounding 8 pixels are numbered 1 to 8 respectively. 0 ”represents each number assigned to a pixel as a digit of each bit by a 9-bit binary number, and the number converted into a decimal number is a pattern number of each pattern. Is obtained (Fig. 3
Specific examples are shown in (B) to (D)).

Further, among the eight surrounding pixels numbered 1 to 8, the flag of the pixel (boundary pixel) whose tracking direction is defined for each pattern is set to "1", and is shown in FIG. 4 (A). As described above, each number assigned to each pixel is represented by an 8-bit binary number and hexadecimal number as a digit of each bit (see FIG. 4).
3B to 3D show specific examples of the pattern numbers shown in FIGS. 3B to 3D). At this time, since the white pixel does not become a boundary pixel, there is no tracking direction, and the tracking direction flags of the pattern (pattern numbers 0 to 255) in which the white pixel is the center are all “00”. Similarly, even if the center is a black pixel, the tracking direction flag is "00" when the black pixel is not a boundary pixel. Then, a table as shown in FIG. 5 showing this tracking direction is stored in the ROM or the like in association with the pattern number.

By defining and storing the tracking direction for each pattern in this manner, the contour tracking of an image can be performed at high speed. When an image as shown in FIG. 6A is input, the stored tracking direction flag is first added to all the pixels of this image, and the tracking direction flag is set to "0".
If the pixels that are "0" are ignored, only the boundary pixels remain (FIG. 7B). Therefore, one of the boundary pixels is used as a starting point, and the tracking direction flag of the boundary pixel serving as the starting point is changed to “00” while moving to a boundary pixel adjacent in either one of two directions. .. Then, since the adjacent boundary pixels have been tracked from one side of the two tracking directions, the tracking is performed in the other direction and the tracking direction flag is changed to "00". Similarly, when the tracking direction flags of the boundary pixels which have been tracked are changed to “00” while tracking is performed according to the direction indicated by the tracking direction flags, all the tracking direction flags are set to “00”. When it becomes, it means that all the boundary pixels are tracked ((C) in the same figure).

Then, some of these tracked boundary pixels are selected and vectorized. As a vectorization method, as shown in FIGS. 7A to 7C, the distance between the boundary pixel and the vector is set as d, and its allowable value is set as dmax in advance, and the distance from the allowable value dmax is set. Boundary pixels that are the bending points of the vector are determined so that d does not become large. For example, when the pixel c is the start point of the vector and the pixel f is the end point of the vector as shown in FIG. 9A, the pixel e that is the distance from the vector among the pixels between the pixel c and the pixel f is The distance d to the vector is d ≦ d
If the distance d is d> dmax when the pixel f is the end point as shown in FIG.
Is set as the end point of the vector, and as shown in FIG.
The pixel which becomes the end point of the vector is searched with using as the start point of the next vector. In this method, the quality of the vector can be set in accordance with the data amount and the required image quality by changing the allowable value dmax.

Next, consider the case where the boundary pixel is a common point. When the boundary pixels of the image as shown in FIG. 8A are extracted, it becomes as shown in FIG. 8B, and the pixel h has both the upper contour line α and the lower contour line β. It becomes a common point to share. At this time, if the tracking directions of the common points are defined in only two directions, only one contour line can be followed. Therefore, four directions of tracking directions are defined for the pixel serving as the common point. At this time, the pixel h and the surrounding pixels are shown in FIG. 9A, and if the surrounding pixels are numbered in the same manner as in FIG. 4A, the tracking direction of the pixel h is pixel numbers 1 and 2. , 4, 5 in four directions. Therefore, the digit flag of this number is set to "1" and an 8-bit binary number and 16
When expressed in a decimal number, it becomes "1B" (hexadecimal). The pattern numbers in FIG. 9A are pixel numbers 1, 2, 4, 5,
Since 9 is a black pixel, it is “283”. Therefore, in this case, the pattern number 2 in the table shown in FIG.
The flag “1B” is stored in the column 83.

However, for example, when the pixel h, which is a common point, is tracked from the direction of the pixel number 4, there are three directions of the pixel numbers 1, 2, and 5, and which direction is followed. In the case of shared points, this table alone is insufficient, because it is not possible to specify whether or not to do so. In particular,
If the tracking directions intersect, vectorization becomes difficult. Therefore, it is necessary to determine the next tracking direction in advance in consideration of the immediately preceding tracking direction so that the contour lines do not intersect. Therefore, when a table as shown in FIG. 10 is separately stored in advance and the pixel h is tracked and the flag of “1B” in the column of the pattern number 283 is designated by the table of FIG. Refer to the table in to determine the tracking direction. In this case, when the pixel is tracked from the direction of the pixel number 1, the pixel is tracked in the direction of the pixel number 5, the pixel number 4 is the pixel number 4, the pixel number 4 is the pixel number 2, and the pixel number 5 is the pixel number 4. By tracking with the pixel number 1, the tracking direction can be tracked without crossing as shown in FIG. 9 (B). In addition, at the common point,
The tracking direction flag is not converted to "00" in the first tracking, but is converted to "00" in the second tracking. Also,
The common point is 1 in total as shown in FIGS.
There are 8 types, and a table similar to that of FIG. 10 is created for each of the 18 types of patterns, and the table of FIG. 5 is referred to. Note that FIG.
(A) to (D), (E) to (H), (I) to (L),
The patterns shown in (M) to (P), (Q) and (R) are objects to be rotated, respectively, and when they are rotated to have the same pattern, the tracking directions are also the same (FIG. 11 (A),
Only (E), (I), (M), and (Q) indicate the tracking direction). FIG. 12A shows an example of the result of tracking the boundary pixels extracted by using the contour tracking method of the binary image described above, and FIG. 12A shows the allowable value dmax of the distance between the boundary pixels and the vector.
FIG. 2B shows an example of vectorization with 1.5 dots (pixels).

[0013]

According to the contour tracking method for a binary image of the present invention, boundary pixels can be tracked by simply applying a predefined tracking direction to an input image, so that it is not necessary to search the tracking direction every time. No, the processing time becomes faster. In addition, since the pixels can be traced without expanding vertically and horizontally by 2 times (handling the data amount of 4 times), there is an effect that a large-capacity memory is unnecessary and the processing time is further shortened.

[Brief description of drawings]

FIG. 1A is a conceptual diagram showing an example of a binary image,
(B) And (C) is a conceptual diagram which shows one Example of the tracking code used for the outline tracking system of the binary image of this invention.

2A to 2F are conceptual diagrams showing another embodiment of the tracking code shown in FIG.

FIG. 3A is a conceptual diagram showing pattern numbers,
(B)-(D) is a conceptual diagram which shows the example.

FIG. 4A is a conceptual diagram for flagging a tracking direction, and FIGS. 4B to 4D are conceptual diagrams showing an example thereof.

FIG. 5 is a diagram showing a correspondence relationship between pattern numbers and tracking direction flags.

FIG. 6A is a conceptual diagram showing an example of a binary image,
(B) is a conceptual diagram showing the contour pixel, and (C) is a conceptual diagram showing the contour pixel traced.

7A to 7C are conceptual diagrams for explaining vectorization.

FIG. 8A is a conceptual diagram showing an example of a binary image, and FIG.
Is a conceptual diagram showing the contour pixel.

9A and 9B are conceptual diagrams showing a tracking direction of a common point.

10 is a diagram showing a correspondence relationship between tracking directions of common points shown in FIG. 9;

11A to 11R are conceptual diagrams showing examples of common points.

FIG. 12A is a conceptual diagram showing a binary image in which all boundary pixels extracted by an embodiment of a binary image contour tracking system according to the present invention are tracked, and FIG. 12B is a vector diagram thereof. It is a conceptual diagram which shows one Example.

[Explanation of symbols]

 a to c, e to h pixels

Claims (1)

[Claims] 1. A contour tracking method for a binary image, wherein some of boundary pixels which are contours of the binary image are extracted and vectorized, and a tracking code corresponding to surrounding pixels is set to pixels of the binary image. A contour tracking method for a binary image, characterized in that vectorization is performed using the boundary pixels obtained by tracking the tracking code.