JPH11134507A - Segment extracting process method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To enable use for the vectorization of various drawing data by finding an upward continuous area paired with a downward continuous area and forming a center line halfway between the both, and finding a leftward continuous area paired with a rightward continuous area and forming a center line halfway between the both. SOLUTION: The outline pixels of a segment are traced in an outline tracing process 1 to acquire a coordinate sequence and it is checked whether or not pixels adjoining each outline pixel in the four upward and downward, and right and left directions are white or black to obtain information. A segment is extracted in an outline segment extracting process 2 from the outline. In a center line extracting process 3, paired segments are detected from segments of the outline area and a center line is found at the intermediate position between both the segments. In a feature point detecting process 4, an intersection and a branch point are found from the center line information. Namely, the upward continuous area paired with the downward continuous area is found to form the center line halfway between the both, and the leftward continuous area paired with the rightward continuous area is found to form the center line halfway between the both.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention detects an intersection, a branch point, an end point, a refraction point, etc. of a line segment from raster data.
The present invention relates to a method for automatically converting a line segment into vector data, which relates to an automatic input tool for digitizing a drawing and a line segment extraction processing method used as a vectorization processing method after extracting a line segment feature from an image.

[0002]

2. Description of the Related Art Conventionally, there is a method of sequentially comparing patterns representing intersections and branch points of line segments with raster data and detecting the data. Accurate detection was difficult with local comparisons.

[0003]

SUMMARY OF THE INVENTION As mentioned above,
Local pattern comparison is greatly affected by distortion and noise, and lacks reliability. Further, since the pattern comparison is sequentially performed on the entire raster data, the processing amount is very large.

An object of the present invention is to provide a high-speed and high-reliability method by solving the above-mentioned problems, and to be able to use the method for vectorization of various drawing data and vectorization after a line detection process from an image. It is an object.

[0005]

When a pixel representing a line segment is tracked, information on a pixel adjacent to each tracking pixel is recorded together with a coordinate sequence representing a trajectory. Based on the adjacent pixel information, a line segment divided by each intersection, branch point, or end point is obtained, and each intersection, branch point, or the like is obtained from the line segment information. In this technique,
The processing amount on the raster data can be reduced by only tracking the outline of the line segment. In addition, in the search for an intersection or the like, more stable detection is enabled by using the position of the line segment, inclination information, and the like.

[0006]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention can be used as a method for automatically recognizing a line segment as vector data from raster data representing the line segment. The following is an embodiment applied to a process of automatically recognizing a line segment from a building plan.

FIG. 1 shows a processing procedure of the present invention. In the drawing, reference numeral 1 denotes contour tracking processing, 2 denotes contour line extraction processing, 3 denotes center line extraction processing, and 4 denotes feature point detection processing. In the contour tracking process 1, the contour pixels of the line segment are tracked, the coordinate sequence is obtained, and the pixels adjacent to each contour pixel in the four directions (up, down, left, right) are checked whether they are white pixels or black pixels. , Have gained that information. Contour line extraction processing 2
In, a line segment is extracted from the outline based on the information obtained in the outline tracking processing 1.

In the center line extraction process 3, a pair of line segments is detected from the line segment of the contour area obtained in the contour line segment extraction process 2, and a center line is obtained at an intermediate position between the two line segments. .
In the feature point detection processing 4, feature points such as intersections and branch points are obtained from the center line information obtained in the center line extraction processing 3.

The contour tracing process 1 will be described. The contour tracking process is a process of sequentially tracking white pixels along black pixels representing a line, for example, as shown in FIG. In this example, tracking of pixels is performed in the vicinity of eight pixels (the next point is selected from eight adjacent pixels).

A specific example will be described below. In order to vectorize a line segment on an image, contour point information of the image is obtained. The trace of the contour point is a white pixel trace (white pixels outside the boundary line with the black region).

FIG. 3 is a diagram for explaining the start point of the outer contour trace and the start point of the inner contour trace. In the figure, a point "untraced white point" is a point to be a trace start point. That is, in the case of the start point of the outer contour trace, when an untraced point that is white is found, it is determined that “a white point is on the point, a white point is on the diagonally upper right, and a black point is on the right. If so, the white untraced point is set as the outer contour trace start point. Similarly, in the case of the start point of the inner contour trace, when an untraced point that is white is found, “the upper part of the point is black, the upper left diagonal is white, and the left part is white. In this case, the white untraced point is set as the inner contour trace start point.

FIGS. 4 (a) and 4 (b) are diagrams for explaining to which next white point the trace is shifted from the current point of interest (determination of the next trace point). FIG.
(A) explains how to read the figure for the "determination condition of the next trace point" shown in FIG. 4 (b). In the figure, the meaning of the symbol indicates a black square as a black point, a white square as a white point, a white circle with a number as a previous trace point, a white circle as a current point of interest, and a diamond as a point that may be white or black. . Further, the “list of trace directions” is a diagram showing the direction as viewed from the current point of interest (white circle). For example, a point in the upper left direction is shown as “1”. And "explanation of figure" is a point of interest (open circle)
When a pattern such as “next-point determination condition” exists under the situation where is placed, the position “7” represented by “next-trace point” is determined as the next trace point, and the trace is performed. To the point.

In FIG. 4B, four cases are given based on the position of the previous trace point, and six patterns are shown in each case. For example, assuming that the previous trace point is the position of “1” or “2”, the situation where the current point of interest (white circle) is placed is “black on top,
When the upper right is white, the right is white, and the lower right, lower, and lower left are arbitrary, the trace is shifted to the white point existing at the position “7” shown in FIG. I do. The next trace point may be considered to be determined “clockwise” with respect to the black point existing around the point of interest (white circle).

A raster scan is performed from the upper left of the image, and a point that satisfies the conditions of the start point shown in FIG. 3 is set as a trace start point. The determination of the next trace is performed for the point of interest under the conditions shown in FIG. Contour tracking is performed based on the above conditions, and at this time, pixel information adjacent to each tracking pixel is also recorded. FIG.
FIG. 5A is a diagram showing that a black pixel representing a line segment is tracked by white pixels (A, B, C, D, E...) Adjacent thereto under the conditions of FIG. In FIG. 5A, it is assumed that the white pixel A extends from the illustrated white pixel α to the white pixel β. Needless to say, the trace ends when returning to the starting point. FIG. 5B shows pixels A, B, C,
Pixels adjacent to each of D and E in the four directions of up, down, left, and right are shown. In this processing, pixels in the four directions adjacent to each pixel of the trajectory are determined by either white or black pixels. Record if there is any. As shown in FIG. 5C, the result is stored in the form of an adjacent pixel table as the coordinates of the pixels in the order of the trajectory and the adjacent pixel information.

The outline segment extraction processing will be described. The following processing is performed on the adjacent pixel information in the four directions obtained by the contour tracking processing. Attention is paid to pixel information in a certain direction, for example, in the upward direction (FIG. 6). The pixel information is checked in the order of the locus, and an area where black pixels are continuous is extracted. Specifically, omission of m pixels is allowed, and a region where black pixels are continuous at n pixels or more is detected. In this processing example, a continuous area is obtained with m = 1 and n = 5.

As a result, the trajectory of the mark A from A 'to B' in FIG. 6 and the trajectory of the mark from C 'to D' are extracted as continuous areas. Regarding the number of pixels m and the number of continuous pixels n that allow this omission, depending on the degree of distortion of the line segment and how many drawings are the minimum units of the line segment inclination,
Change and respond. The method of determining the numerical value is automatically determined according to conditions such as how a person judges and tunes a value that gives the most appropriate result for the sample drawing and how much the line segment can be distorted. It corresponds to the drawing and the adaptation example, such as the method of determining.

FIG. 6 shows an example in which a process for searching for a place where black exists in the upper direction is performed. Similarly, a place where black exists in the downward direction, a place where black exists in the left direction, and a right direction Like where there is black,
After processing is performed for each of the upper, lower, left, and right directions, the extracted regions are overlapped. At this time, a portion that does not overlap with the continuous region in the other direction is separated from a portion that overlaps pixels in the upper and right, upper and left, lower and right, and lower and left directions. As a result,
Eight types of continuous areas, lower, left, right, upper and right, upper and left, lower and right, and lower and left, are obtained. In the embodiment of FIG.
It is divided into six types of continuous areas: left, right, upper & right (upward continuous area and rightward continuous area), and lower & left. In FIG. 7, upper & left and lower & right do not appear.

The center line extraction processing will be described. As shown in FIG. 8, the present process finds a pair of continuous regions obtained as described above, for example, a pair of a downward continuous region and an upward continuous region, A center line serving as a line segment candidate is obtained at an intermediate position between the two. At this time, line segment thickness information is also acquired. The paired continuous areas are opposing contour lines forming a line segment. This pair of continuous regions is
I think it always exists in the downward direction. Similarly, right on the left,
Search by the combination of bottom and left on top and right, and bottom and right on top and left. By limiting the search area in this way, the amount of calculation is reduced, and a portion where the center line position near the branch point or the intersection as shown in FIG. 9 is likely to be shifted can be excluded. In FIG. 9, when the center line is obtained by the conventional method and the intersection of the center line is obtained, the position of the intersection of the center line is changed to the original position p as shown by the “conventional method”.
However, as shown in “this method”, it indicates that the displacement is unlikely to occur.

A specific example of searching for a pair of continuous areas will be described below. Construct a list of start and end points for the continuous area in the upward direction. Next, the slope of the downward continuous area is calculated from the start and end points of the downward continuous area, and the respective upward and downward continuous areas are searched from each of the start and end points of the downward continuous area in a direction perpendicular to the slope. Check if there is. The range to be searched in the vertical direction is set until a white pixel appears for the first time or within a certain value obtained based on the maximum line segment width in the drawing.

When an upward continuous area exists, the start and end point information of the downward continuous area is registered together with the corresponding position in the start and end point list of the upward continuous area. Also, the distance from the search start point to the search end position is secured as line segment thickness information. In the case where the start and end points in both the upper and lower directions are searched in the registered list in the right direction, for example, as shown in FIG. 8, the start point of the upward continuous area is followed by the start point of the downward continuous area In, find the overlapping portion and generate a center line at the intermediate position. Similar processing is performed for each corresponding continuous area.

The feature point detection processing will be described. In this process, in order to obtain a feature point such as an intersection or a branch point, first, the center line generated earlier is classified for each forming the same feature point. After that, the position of the feature point is determined. Specific processing will be described with respect to the characteristic points shown in FIG.

The trajectory B is used to generate the center line 1. Therefore, from the pixel on the trajectory B that generated the end point 1, the trajectory is searched in a direction other than the contour of the center line 1 up to the "pixels that are contours of other center lines". In this example, since a pixel which is the outline of the center line 2 is found, it is understood at this point that the end point 1 of the center line 1 and the end point 2 of the center line 2 are connected by the same feature point. Similarly, it can be seen from the other trajectory C that generated the center line 2 that the center line 2 is connected to the center line 3. By repeating this process, it can be seen that one feature point is formed by the center lines 1 to 4. The above processing is performed for all the center line end points, and the connection relation of all the feature points is obtained.

The method of determining the position of a feature point is to extend the end points so that they are closest to each other, to determine the position preferentially from the position of the long center line, or to preferentially consider horizontal and vertical line segments. Various methods are conceivable, such as a method of determining the position by using the method.

An example of preferentially using the position of a line segment in the horizontal and vertical directions will be described below. The one whose center line is horizontal / vertical within a certain allowable value is selected. Among them, line segments having the same feature point are determined as continuous line segments. After determining a continuous line segment, if it is a horizontal line, the y coordinate is
In the case of a vertical line, the coordinate value of the line segment is corrected using the average value of the x coordinate as a new coordinate. At this time, the line segment whose coordinates are to be corrected is only a single line segment, and when there are two or more connected line segments (double line or the like), the end of a continuous line segment is determined.
Exclude from the target of coordinate correction. This coordinate value correction processing is shown in FIG. The coordinate correction is performed on the single line surrounded by the dotted line in the center diagram of FIG. 11, and the y coordinate is corrected as shown in the lower diagram of FIG.

Next, a connection process at a characteristic point is performed on the line segment for which the coordinate correction of the horizontal and vertical lines has been completed (FIG. 12). The line segment is forcibly extended and connected. For feature points that have horizontal and vertical segments, the intersection point is set to the coordinates where the respective segments are extended.If only the horizontal segment is extended or only the vertical segment is extended, the average coordinates of the midpoint I do.
If horizontal and vertical line segments and oblique lines are mixed in the same group (FIG. 13), the end points of the oblique lines are the intersection coordinates of the horizontal and vertical lines. At this time, if there are two or more parallel lines in the horizontal and vertical line segments, there are a plurality of intersection coordinates, so the end point coordinates of the oblique lines are the average coordinates of the horizontal and vertical line end points.
If only one horizontal line or one vertical line exists in the same group, the end point has not been determined, and the coordinates of the oblique line are the intersections with the line segment. At this time, the end point of the horizontal line or the vertical line is also changed to the intersection coordinates. The feature points are
In the case of only the diagonal lines (FIG. 14), the intersection coordinates obtained by combining two lines each are registered as candidates. At this time, if the opening angle of the line segment is close to 180 degrees, the coordinates of the middle point are used as an exception. Finally, the average value of the coordinates registered as candidates is taken as new coordinates.

The present invention also includes a storage medium in which a program for executing the above-described line segment extraction processing method is stored on the storage medium.

[0027]

As described above, according to the present invention, the outline of a figure is tracked from inside and outside, and black pixels existing above, below, right or left as viewed from the white pixel of interest continue. Since a continuous region is examined and a center line is generated in the middle of two continuous regions that exist as a pair, it is less affected by local distortion and noise, and at intersections and branches. Feature points such as points can be determined stably.

[Brief description of the drawings]

FIG. 1 is a diagram showing a processing flow of the present invention.

FIG. 2 is a diagram illustrating contour tracking.

FIG. 3 is a diagram showing setting of a trace start point.

FIG. 4 is a diagram illustrating determination of a next trace point.

FIG. 5 is a diagram illustrating collection of adjacent pixel data during contour tracking.

FIG. 6 is a diagram showing pixels in an upward direction.

FIG. 7 is a diagram showing a continuous area for each direction.

FIG. 8 is a diagram showing a pair of continuous regions and a center line.

FIG. 9 is a diagram showing a center line comparison with a conventional method.

FIG. 10 is a diagram showing a locus and a center line.

FIG. 11 is a diagram showing correction of coordinates of horizontal and vertical line segments.

FIG. 12 is a diagram showing connections of horizontal and vertical line segments.

FIG. 13 is a diagram showing connection of oblique lines including horizontal and vertical line segments.

FIG. 14 is a diagram showing connection of only diagonal lines.

[Explanation of symbols]

 1: Contour tracking processing 2: Contour line extraction processing 3: Center line extraction processing 4: Feature point detection processing

Claims (5)

[Claims] 1. A line segment extraction processing method for extracting a line segment constituting a figure using a contour of the figure, wherein a white pixel in contact with a black pixel of a given figure is tracked, and the individual white pixel is traced. Is stored as an adjacent pixel table about which pixels around the white pixel are white pixels and which pixels are black pixels, and based on the contents of the adjacent pixel table, the black pixels existing in the upward direction Are present in the left direction and an upward continuous region consisting of a white pixel group in which a continuous To extract a leftward continuous region consisting of a white pixel group in which black pixels are continuous and a rightward continuous region consisting of a white pixel group in which black pixels present in the right direction are continuous, Upward continuation region that is paired with downward continuation region A line characterized by finding a region and forming a center line in the middle between the two, and finding a left continuous region paired with the right continuous region and forming a center line in the middle between the two. Minute extraction processing method. 2. In order to detect a feature point including an intersection and a branch point on the basis of information on each of the formed center lines, the starting point coordinates and / or the end point coordinates of each of the formed center lines are determined. Based on this, the center lines that would form the same feature point are grouped for each individual feature point, and then the grouped center lines are connected, and each feature point is connected to the connected portion. 2. The line segment extraction processing method according to claim 1, wherein it is determined that the line segment exists. 3. In connecting the center line groups to determine a feature point, a priority is given to the center line of a predetermined pattern in the center line group, and the center line on the given center line is given the priority order. In determining the location of the feature point, a point or a point on the extension of the center line is determined temporarily as the location of the feature point in preference to other center lines. The line segment extraction processing method according to claim 2. 4. The system according to claim 3, wherein the center line given the priority is an intersecting line segment in the horizontal and vertical directions.
The described line segment extraction processing method. 5. A storage medium storing a line segment extraction processing method for extracting a line segment constituting a figure using a contour of the figure, wherein a white pixel in contact with a black pixel of a given figure is tracked, For each individual white pixel, which pixel around the white pixel is a white pixel and which pixel is a black pixel is stored as an adjacent pixel table, and exists in an upward direction based on the contents of the adjacent pixel table. An upward continuous area consisting of a group of white pixels having continuous black pixels, a downward continuous area consisting of a group of white pixels having continuous black pixels present in a downward direction, and a leftward direction A continuous region in the left direction consisting of a group of white pixels in which black pixels existing in the right direction is continuous, and a continuous region in the right direction consisting of a white pixel group in which black pixels existing in the right direction are continuous. So that it can be extracted A line that finds an upper continuous region and forms a center line in the middle between the two, and finds a left continuous region paired with the right continuous region and forms a center line in the middle between the two. A storage medium storing a minute extraction processing method.