JPH0277887A - Contour extraction system for binary graphic form with direction code - Google Patents

Abstract

PURPOSE:To extract all contours in a graphic figure whatever shape the binary graphic form is in by checking the marking state of a pair of white and black picture elements and extracting even a black picture element which is retrieved once as the extraction start point of a next contour. CONSTITUTION:When the extraction start point (SX,YX) of a contour Ei is determined for contour extraction, the direction codes of respective black picture element points are extracted in order while a mask is setting at respective black picture element points constituting contour points. Consequently, the direction code point sequence (shown in figure) of connecting contours Ei is extracted automatically and eventually. Thus, the extraction of the direction code point sequence of the contours Ei is finished, the contour lines are decided from a decision making table with a tail direction code iD=7 and the extracting operation for contours E0 and Ei of the binary graphic pattern is finished.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to a binary figure outline extraction method for extracting the outline of a binary figure using an 8-connected direction code [Prior art] When extracting a shape or calculating its feature values such as its center of gravity and perimeter, a binary figure is sometimes converted into contour information for ease of handling. 2 like this
One method of describing the contour of a value figure is to use a so-called 8-connected direction code iD consisting of eight directions (up, down, left, right, and diagonal) as shown in FIG. 11, and to describe it by a sequence of points of this direction code. For example, a binary figure P having the shape shown in FIG.
In the case of , the circumferential contour is the direction code point sequence (6°5,
4.6.0,7,6.0.2.2,2,2,2.4).

In order to extract the outline of a binary figure as an 8-connected direction code in this way, prepare a mask M consisting of 9 squares of 3 x 3 pixel size as shown in FIG. As shown in the figure, it is set on a binary figure, and eight squares M around the eyes are set at the center. - The black pixel that provides the next contour point by searching the black and white of each pixel located at M7 counterclockwise (in the case of Figure 14, the black pixel directly below)
is detected, the direction code iD (in the case of FIG. 14, 1D=6) giving the direction of movement to this black pixel point is obtained, the mask M is moved to the black pixel point, and the same process is repeated. As shown in Figure 12, contour information consisting of a series of direction code points was extracted.

[Problem to be solved by the invention]

When extracting the outline of a binary figure, it is necessary to extract all the sharp points of the figure, lines with a width of 1 pixel, as well as the internal and external lines of the figure, faithfully and at high speed. However, the conventional method is still not satisfactory in terms of processing speed, and black pixel points extracted as contour points cannot be used as starting points for extraction of another contour, and depending on the figure shape, all In some cases, it was difficult to extract the contour of the image. Furthermore, it is difficult to know whether the extracted contour is the inner line or the outer line of the binary figure (also, the contour of the binary figure is extracted using an 8-connected direction code). In this case, for example, 2 shown in FIG.
At a concave right angle point in the value shape, its direction code is (6, 5
,4,...0,7,6. ...), resulting in an 8-connected state, which had the disadvantage that right angle points could not be extracted faithfully.

The present invention was invented in view of the above circumstances, and an object of the present invention is to provide a contour extraction method that can extract all contours of a binary figure quickly and accurately.

[Failure to solve the problem]

In order to achieve the above object, the invention described in claim (1) first searches for the black and white state of eight pixels around the eye in a counterclockwise direction from the center of the mask, and A black pixel point is marked, and the next contour extraction start point is determined by referring to the marking information in units of adjacent white and black pixels.

Furthermore, the invention described in claim (2) provides a comparison table that determines the optimum search start direction for a black pixel point that provides the next contour point for the previous direction code, and by referring to the comparison table. The optimal search start direction of a black pixel point that gives the next contour point is determined from the previous direction code, and the black and white state of the pixel is sequentially searched counterclockwise from the pixel at the position of the optimal search start direction.

Further, the invention described in claim (3) provides a determination table for determining whether the contour is an internal line or an external line from the final direction code of the contour, and by referring to the determination table, the contour is Whether the contour is an internal line or an external line is determined from the final direction code.

Further, the invention described in claim (4) provides a conversion table for converting an 8-connection direction code into a 4-connection direction code, and by referring to the conversion table, the 8-connection direction code is converted into a 4-connection direction code. The direction code is converted to the following direction code.

[For production]

In the case of the invention described in claim (1), the black pixel at the extraction start point of the next contour is detected while referring to the marking state of adjacent black and white pixels as a pair. Therefore, by looking at the marking state of the black and white pixels forming this pair, especially the marking state of the white pixel, even a black pixel point that was searched as a -degree contour point can be picked up again as a starting point for extraction of the next contour. Can be done.

In the case of the invention according to claim (2), a black pixel point that provides the next contour point is sequentially searched using the pixel at the optimum search start direction position obtained by referring to the comparison table as a base point.

Therefore, the search for a black pixel point that gives the next contour becomes faster.

In the case of the invention described in claim (3), it is determined whether the contour is an inner line or an outer line of a binary figure based on the final direction code of the contour with reference to the determination table. Therefore, at the same time as the contour is extracted, it is possible to automatically know whether the contour is an internal line or an external line of a binary figure.

In the case of the invention described in claim (4), an 8-connection direction code can be converted into a 4-connection direction code by referring to a conversion table. Therefore, the nature of the direction code concatenation (
8-connection/4-connection) can be used to collect contour information of a binary figure, making it possible to extract the contour faithfully.

〔Example〕

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 2 is a block diagram showing one embodiment of a contour extraction device to which the method of the present invention is applied, and FIG. 1 is a flowchart of its operation.

In FIG. 2, reference numeral 1 denotes a scanner that captures a binary figure P having an arbitrary shape as image data of a dot image; 2 a figure memory consisting of a RAM that stores the image data of the binary figure captured from the scanner 1; 3 is a table memory consisting of a ROM that stores direction codes and various reference tables to be described later; 5 is a contour point string memory consisting of a RAM that stores a direction code point string of a binary figure whose contours have been extracted; and 6 is an operation of the device. This is a control unit consisting of a CPU and the like.

The table memory 3 includes a comparison table that determines the optimal search start direction for a black pixel point that provides the next contour point for the previous direction code iD shown in FIG. 3, and a final direction of the contour shown in FIG. A determination table that determines whether the contour is an internal line or an external line based on the code iD, and a conversion table that converts the 8-connected direction code iD shown in FIG. 5 into a 4-connected direction code F4 (iD). Furthermore, the 8-connection direction code iD shown in FIG. 11 is stored in advance.

Proceeding, with reference to the flowchart of FIG. 1, the outline E of the binary figure P shown in FIG. The operation will be explained by taking as an example the case where and E are extracted as direction codes.

In the following explanation, white pixels for which the search has not been completed are indicated by "mouth", white pixels for which the search has been completed and marked are indicated by "・", and black pixels for which the search has not been completed are indicated by ".". A black pixel that has been marked after the search is indicated by a "■" mark, and a marked black pixel after the search is completed is indicated by a "Congratulations" mark. . However, in the drawings, in order to prevent adjacent black pixels from becoming indistinguishable, black pixels for which the search has not been completed are indicated by cross-hatching with diagonal lines on the left and right instead of the above-mentioned "■".

A binary figure stored in the figure memory 2 is scanned to find one of the black and white pixel pairs ``mouth ■'', ``four places'', or ``•■'' that first collides with the scan line. Then, the outline extraction starting point (
sx, yx). In the case of FIG. 6, the black and white pixel pair that first collides with the scan line is the black and white pixel pair "mouth ■" located at the top of the left outer surface, so the black pixel point of this black and white pixel bare is contour E. extraction starting point (SX,
YX) (steps [1] [2]).

Contour E as described above. extraction starting point (SX.

When YX) is determined, the search start direction code FN is set to the initial value FN=5N, and the black pixel point that gives this extraction start point (SX, YX) is marked,
Change it to "厘" indicating the end of the search (step [3]).

Next, as shown in FIG. 7(a), a mask M (11th
First, match the eyes and set the search start direction F.
Starting from the position of the eye M, which corresponds to N=5, that is, the direction code 1D=5 of the mask M, the black and white state of each pixel covered by the eight squares around the center square is rotated counterclockwise (reversely). (clockwise) (steps [5] [6]
). At this time, the white pixel point for which the search has been completed is marked to indicate the end of the search, changing from "mouth" to ".".

In the case of FIG. 7(a), the first black pixel is detected at the position of square M6. When this black pixel is detected, after confirming that the detected black pixel point does not match the extraction start point (sx, yx) (step "8J"), the black pixel point is assigned to the next contour point. The direction code 1D of the black pixel point is determined as the pixel position, and the center of the mask M is viewed from the eye as the reference point.
=6 is contour E. is stored in the contour point sequence memory 5 as the first direction code (step [91). Then, from the comparison table in FIG. 3, the optimum search start direction FN=3 for the next black pixel point when the direction code iD-6 is determined, and the searched black pixel point is marked and placed in a "cage". (step [101[11]).

Next, as shown in FIG. 7(b), the mask M is moved to the position of the black pixel point with the detected direction code 1D=6, and the optimum search start direction FN=3, obtained from the comparison table, is set. That is, using the position of the square M3 corresponding to the direction code 1D=3 of the mask M as a base point, the next black pixel point is searched counterclockwise in the same manner as described above.

By performing the above processing one after another for each black pixel point composing the contour, the final contour E shown in FIG. 6 is formed. The direction code point sequence for (6,
6,5,6,6,6,6,2,1,0,0,0,0,0
,0゜0.2.2,2.2,2,2.2,4,4,4,
4, 4, 4.4) are automatically extracted (step [1
2] [13]).

When the extraction of the direction code point sequence of the contour E0 is completed, the final direction code 1D=4 determines from the determination table in FIG. (Step [14]).

That is, in the case of a two-dimensional binary figure, the extracted contour is described by a series of points starting from the extraction starting point and returning to the extraction starting point again. When expressing a contour using a direction code, the final direction code returning to the extraction starting point is 1 due to the nature of binary figures.
D=2,3.

There are only 4 possible directions of 4.7, and 1D=2.3
．． 4, the contour is the outer line of the figure, and is 1D
When =7, the contour is uniquely determined to represent the internal line of the figure. Therefore, by preparing in advance the final direction code iD and the distinction between internal and external shadow lines as shown in FIG. It is possible to automatically determine whether it is an outline line.

Contour E showing the external line as described above. When the extraction of E is completed, the process moves on to extracting the contour E that constitutes the inner shape line.

In this case, as in the case described above, the binary figure in the figure memory 2 is raster-scanned, and the black and white pixel bear 10 that first collides with the scan line is
Find one. In this case, the black and white pixel bear "low" located at the top right side of the internal space of the binary figure P is selected, and the black pixel point of this pair is the extraction start point (SX, Y
X). Note that at this time, the above-mentioned contour E has been extracted. The black and white pixel bears located on the contour line also collide with the raster scan line one after another, but both of the black and white pixels of these black and white pixel bears fall on the contour line E. It is marked at the time of extraction, and is a combination of "-". Therefore, these black and white pixel pairs are never selected as the starting point for extraction of the contour E8.

As described above, the extraction start point (SX) of the contour E.

Once YX) is determined, the contour E described above is determined.

As in the case of extraction, a mask M is set at each black pixel point constituting the contour point, and the direction code of each black pixel point is sequentially extracted. As a result, the direction code point sequence (6, 6, 6, 6,
5,4,4,4,3,2,2,2,l, 0.0,0.
7) is automatically extracted. Note that the rotation direction of the contour E that provides the inner line is the contour E that provides the outer line.

It goes in the opposite direction.

When the extraction of the direction code point sequence of the contour E is completed, the final direction code 1D=7 indicates that the contour E is a contour line that provides the inner line of the binary figure from the determination table in FIG. is determined (step [14] L). The outline E of the binary figure P shown in FIG. 6 is determined. Then, all extraction operations of Ei are completed.

FIG. 8 shows another example of contour extraction. This example is shown in the same figure (a
), this is an example in which a binary figure is composed of two independent figure parts "mouth" and "I". The contour extraction process is shown in the same figure).

Contour E is performed in the order of (C) and (d). , -(6,
6,6,0,0,0゜0.0,2,2.2,4,4,4
,4.4),Contour E. 2 = (6, 6, 6, 2, 2゜2
), contour E knee (6, 5, 4, 4, 3, 2, 1, 0, 0
,7).

By the way, when extracting the outline of a binary figure using the above-mentioned 8-connected direction code iD (Fig. 11), the direction code is bent at right angles at the contour parts where it is bent at right angles and becomes concave. Instead of progressing, it is extracted as a diagonal direction code, resulting in a condition called 8-concatenation.

For example, in the case of the contour E extracted in Fig. 6 above, Ei
-(6, 6, 6, 6, 5, 4, 4, 4, 3, 2, 2,
The underlined portions (21, 0, 0, 0°7) result in 8 connections, and the concave right angle portions cannot be extracted faithfully.

Therefore, in the present invention, in order to prevent this, a conversion table for converting an 8-connection direction code into a 4-connection direction code is prepared as shown in FIG.

Based on this conversion table, if the 8-connected direction code is converted to the 4-connected direction code F4 (iD) at the timing positions of steps [9 and 12] in FIG. 1, the concave right angle It is possible to prevent the so-called 8-connection state from occurring at some locations.

This conversion operation in FIG. 5 is performed by converting the diagonal direction code iD =
Adjacent left and right horizontal and vertical direction codes i of L3+5+7
This is a process of replacing it with either D-0 or 2+4+6. By performing this conversion, the 8-connected direction code i
D = O+1+2+3+4+5+6+7 is converted into a 4-connected direction code F 4 (i D) -0, 2, 4, 6.

FIG. 10 shows an example in which the contour of a binary figure extracted by an 8-connected direction code is converted into a contour of a 4-connected direction code using the conversion table. In the figure, ○ indicates 8 connection points, and "." indicates 4 connection points. As is clear from the figure, by converting to 4-connection, the 8-connection state is eliminated, and the concave right angle portion is faithfully extracted.

〔Effect of the invention〕

According to the invention described in claim (1), even a black pixel that has been once searched can be extracted again as a starting point for extraction of the next contour by looking at the marking state of the black and white pixels forming a pair, so that a binary figure can be extracted. It is possible to extract all contours in a figure, no matter what shape it is.

According to the invention set forth in claim (2), since the black pixel points that give the next contour point are sequentially searched based on the optimal search start direction, the extraction time for the black pixel points that give the next contour point is shortened. This makes it possible to speed up the processing.

According to the invention set forth in claim (3), since it is determined from the final direction code of the contour whether the contour is an inner line or an outer line of a binary figure, it is possible to determine whether the contour extraction is finished. At the same time, the type of contour can be automatically determined.

According to the invention described in claim (4), since the 8-connected direction code is converted to the 4-connected direction code,
It becomes possible to collect contour information of a binary figure by specifying the nature of the connection of the direction code (8-connection/4-connection), and it becomes possible to faithfully extract the shape of the binary figure.

[Brief explanation of the drawing]

Figure 1 is a flowchart for explaining the operation of the method of the present invention, Figure 2 is a block diagram showing an embodiment of the method of the present invention, and Figure 3 is a black pixel that provides the next contour point from the previous direction code. Figure 4 is a diagram showing an example of a comparison table for determining the optimal search start direction for a point. Figure 5 is a diagram showing an example of a conversion table for converting an 8-connected direction code into a 4-connected direction code, Figure 6 is a diagram showing an example of outline extraction of a binary figure according to the present invention, and Figure 7 is a 3× FIG. 6 is a process explanatory diagram when contour extraction is performed according to the present invention by setting a 3-pixel size mask, FIG. 8 is a diagram showing another example of contour extraction of a binary figure according to the present invention, and FIG. 9 is a diagram showing another example of contour extraction of a binary figure according to the present invention An explanatory diagram of conversion from an 8-connected direction code to a 4-connected direction code, Figure 1O shows the contour extracted by the 8-connected direction code
A diagram showing an example of the case where the direction code of connection is converted to the outline. Figure 11 is a diagram showing the direction code of 8 connections, and Figure 12 is 8.
An explanatory diagram of the conventional outline extraction of binary figures using the direction code of connection. Figure 13 is a diagram showing a mask of 3 x 3 pixel size, Figure 14
The figure is an explanatory diagram of the principle of contour extraction using a mask of 3×3 pixel size. P... Binary figure, M... 3x3 pixel size mask, iD... 8-connected direction code, FN (iD).
... Optimal search start direction, F4 (iD) ... 4-linked direction code. Patent applicant Rico Co., Ltd.

Claims (4)

[Claims] (1) Using raster scanning, first detect the black pixel point that changes from a white pixel to a black pixel and use it as the starting point for contour extraction, and set a 3x3 pixel size mask on the black pixel points that make up the contour point. By sequentially searching the black and white states of pixels covered by the squares of the mask in a counterclockwise direction, the direction code of the black pixel point that gives the next contour point is detected, and the contour of the binary figure is determined by the point sequence of the direction code. In the contour extraction method using the direction code of a binary figure, which is expressed as A contour extraction method using a direction code of a binary figure, which is characterized in that the next contour extraction start point is determined while the extraction start point of the next contour is determined. (2) Prepare a comparison table that determines the optimal search start direction for the black pixel point that gives the next contour point for the previous direction code, and by referring to the comparison table, search for the next contour point from the previous direction code. 2 according to claim (1), characterized in that an optimal search start direction of a black pixel point giving a contour point is determined, and the black and white state of the pixel is sequentially searched counterclockwise using the pixel at the position of the optimal search start direction as a base point. Contour extraction method using direction code of value figure. (3) Prepare a judgment table that determines whether the contour is an internal line or an external line from the final direction code of the contour, and by referring to the judgment table, it is possible to determine whether the contour is an internal line based on the final direction code of the contour. 2 according to claim (1), characterized in that it is determined whether the line is a shape line or an outline line.
Contour extraction method using direction code of value figure. (4) A conversion table for converting an 8-connection direction code to a 4-connection direction code is prepared, and the 8-connection direction code is converted to a 4-connection direction code by referring to the conversion table. A contour extraction method using a direction code of a binary figure according to claim (1).