JPH0460879A - Detecting method for feature point position and straight line direction on boundary line - Google Patents

Abstract

PURPOSE:To inspect the uniformity of a pattern in respect of the picture of a metallographic structure cross section, etc., by detecting the position of a feature point and the directional relation of each straight line at the time when two areas different in their brightness are adjacent to each other as considering each straight line extending in two different direction from the feature point to be a boundary by picture processing. CONSTITUTION:A window W is generated in a fixed picture including the feature point P in a field of view so that one of the straight lines penetrates it in parallel with its one side, and at every place at the time when this generated window is moved in parallel with one of the straight line, a projection upon a first side perpendicular to the moving direction of the window is taken. Then, the coordinate of the changing point of the maximum value or the minimum value of this projection in relation to the first side of the window is determined for every place, and the position of the feature point is detected based on two pieces, at least, of this coordinate. Besides, the projection upon the second side of the window perpendicular to the first side is taken for every place, and the directional relation of each straight line is detected based on the increasing or decreasing tendency of this projection value. Thus, this method can be utilized for the automatic inspection of the uniformity of the pattern in respect of the picture of textiles or the metallographic structure cross section, etc.

Description

[Detailed description of the invention] [Industrial application field]

The present invention is a method that can be used to inspect the pattern (pattern) of an image of, for example, a cross section of a fabric or a metallographic structure, in which two areas of different brightness are located in two different directions from a single point as a feature point. The present invention relates to a method for detecting the position of a feature point on a boundary line and the direction of a straight line, in which the position of the feature point and the directional relationship of each straight line are detected by image processing based on imaging when adjacent feature points are adjacent to each other with each line extending as a boundary.

[Conventional technology]

When the correct pattern is for two areas of different colors or brightness to be adjacent to each other with a straight line as their boundary, suppose that three areas now have a straight line extending from one point in two different directions as their boundaries. The - point at that time is called a feature point, and the conventional method of detecting the position of this feature point and the directional relationship between each straight line in two directions is to image the range that includes the feature point and use it as a CR.
An image was displayed on a T-display, and detection was performed by visually observing this image.

[Problem to be solved by the invention]

As explained above, in the conventional technology, the position of the feature point and the directional relationship of each straight line in two directions are detected by visually observing the image on the CRT display. Not only does it lack stability, but it also poses problems in terms of work management, such as fatigue and securing personnel. The object of this invention is to solve the problems of the conventional techniques, and to make it possible to automatically inspect the characteristics of patterns on images of textiles, metal structure cross sections, etc. The object of the present invention is to provide a method for detecting point positions and linear directions.

[Means to solve the problem]

In order to solve this problem, the feature point position and
The linear direction detection method is based on the following: When two areas with different brightness are adjacent to each other with straight lines extending in two different directions from one point as a feature point as boundaries, the position of the feature point and the directional relationship of each straight line are determined. In a method of detecting by image processing based on imaging, a rectangular window is generated in both fixed images including the feature point in the field of view, such that one of the straight lines passes through it parallel to one side of the rectangular window. At each point when the window is moved parallel to the one straight line, take a projection onto the first side perpendicular to the direction of movement of the window, and determine the starting point of change of the maximum or minimum value of this projection. determining the coordinates of the first side of the window for each of the locations; detecting the position of the feature point based on the coordinates of each location; Steps of: taking a projection onto the second side for each location and determining its projection value; detecting the directional relationship of each straight line based on the tendency of increase/decrease in the projection value for each location; Be prepared.

[Effect]

A rectangular window is generated in both fixed images that include the feature point in the field of view, with one of the straight lines passing through it parallel to one side of the window, and this generated window is moved parallel to one of the straight lines. At each point in time, a projection is taken onto the first side perpendicular to the moving direction of the window. Then, the coordinates of the starting point of the change in the maximum or minimum value of this projection, relative to the first side of the window, are determined for each location, and the position of the feature point is detected based on at least two of these coordinates. Further, the projection onto the second side perpendicular to the first side of the window is taken for each of the above locations, and the directional relationship of each straight line is detected based on the increase M (lJi direction) of this projection value.

[Example] An application example of the method for detecting the feature point position and linear direction of the boundary line F according to the present invention will be described below with reference to a cross section. FIG. 1 is a schematic diagram showing the relationship between a fixed image and a window. In FIG. 1, within a fixed image field F obtained by imaging the target pattern, two lines are defined by straight lines Ll and L2 extending upward and diagonally downward to the right from the feature point P, respectively. There is an area DLD2. Now, area D
1 is dark and area D2 is bright. 1. Straight parallel to one side. ! When a rectangular window W through which L1 passes is generated and it is moved in the direction of the straight line L1, the projection onto the X and Y axes at each location described below is determined. The size of this window W is, for example, such that a plurality of windows W can be accommodated within the fixed image field of view F, and the X-direction component of the straight line L2. Make it twice or less of each value of the X direction component. FIG. 2 is a schematic diagram showing four positions of the moving window. In the figure, the windows at each of the four positions are shown as Wl, W2, . W3. Let's call it W4. The feature point P is not included in each window Wl, W4, but is included in each window W
2. Included in W3. The X-axis and Y-axis are parallel to the upper and left sides of the fixed image field of view F, respectively.
Take the axis. The coordinates of the feature point P are Xo, Yo, each window W2. When the intersections of the lower side of W3 and straight line L2 are points A and B, the coordinates of each point A;B are XLYI;
2. Let it be Y2. Figure 3 is a characteristic diagram of the projection value in the X-axis direction with respect to the
is a characteristic diagram corresponding to the window W2, (C) is a characteristic diagram corresponding to the window W3, and (d) is a characteristic diagram corresponding to the window W4. In FIG. 3(a) to (d), the horizontal axis is the X coordinate,
The projection value 3× onto the X axis is plotted on the vertical axis. Figure (a)
Then, at the coordinate Xo, the projection value Sx rises steeply from a low position to a high position. Similarly, in the same figure (b), the coordinate χ0
The projection value Sx begins to increase from a low position and becomes flat at a high position from the coordinate X1. In the same figure (C), the projection value Sx starts to increase from a low position at the coordinate Xo and becomes flat from a high position at the coordinate X2. In the figure (d), the projection value Sx is low and flat. Here, the low and high values of the projection value Sx correspond to darkness and brightness of each area DI and D2, respectively. By the way, it should be noted that the projection values on each coordinate axis vary considerably depending on the unique characteristics of the image and slight differences in the way the illumination is applied. Therefore, it is necessary to detect the projection value in such a way that it changes extremely depending on the presence or absence of the factor. Based on the coordinates Xo, XlX2 obtained from Figure 3 (a), (b), and (C), and the coordinates Yl, Y2 in Figure 2, the coordinates are calculated based on the formula for a straight line connecting two points in analytical geometry. Y. is required. Yo =Y1+(Y2-Yl)(Xo-X'l)/(
X2 - XI) In the above manner, the position of the feature point P has been detected by obtaining the respective coordinates Xo and Yo. In addition, in FIG. 2, if the window W is moved in small increments and the number of windows containing the feature point P is three or more, the position of the feature point P can be detected more accurately than described above. Next, a method for detecting the directional relationship between the straight lines LLL2 extending in two directions from the feature point P will be described with reference to FIG. Figure 4 is a characteristic diagram of the projection value in the X-axis direction versus the Y-coordinate at four locations of the moving window, where (a) is a characteristic diagram corresponding to window W1 in Figure 2, and (b) is the characteristic diagram corresponding to window W1 in Figure 2.
is a characteristic diagram corresponding to window W2, II (C) is a characteristic diagram corresponding to window W3, and FIG. 4 (j) is a characteristic diagram corresponding to window W4. In common, the vertical axis is the Y coordinate,
The horizontal axis represents the projection value Sy onto the Y axis. Figure (a)
In this case, the projection value Sy takes a slightly high, almost constant value. In FIG. 4B, the projection value Sy shows a decreasing tendency as the Y coordinate increases after passing the initial stage of the Y coordinate. In the figure (C), the projection value Sy shows a decreasing tendency as the Y coordinate increases after passing an initial stage that is later than in the figure (b) of the Y coordinate. In the same figure (cl), the projection value Sy takes a low, almost constant value. In Fig. 4(b) and (C), the coordinate Y of the projection value Sy
The direction of the straight line L2 in FIG. 2 can be determined from the decreasing tendency for the curve. In Fig. 2, if the direction of the straight line L2 is indicated by a broken line, then in Figs. In FIG. 4(d), the projection value Sy takes on a high, almost constant value as indicated by the broken line. In this manner, the direction of the straight line L2 indicated by the broken line in FIG. 2 can be determined from the upward and downward tendency of the projection value sy with respect to the coordinate Y. Returning to FIG. 1 again, when there is a weaving error in the fabric to be inspected, each coordinate Xo, Y
o and the directional relationship between the straight lines Ll and L2. In addition, when the pattern is −-like, each straight line Ll,
L2 becomes the same line, and the feature point P does not exist. Effects of the Invention As explained above, according to the present invention, when two areas with different brightness are adjacent to each other with straight lines extending from one point as a feature point in two different directions as boundaries, the feature point The position of the line and the directional relationship of each straight line can be detected by image processing based on imaging, and this detection method can automatically inspect the pattern of images of textiles, metal structure cross sections, etc. It can be used to improve inspection accuracy, efficiency, and labor savings.

[Brief explanation of drawings]

Figure 1 is a schematic diagram showing the relationship between the fixed image and the window, Figure 2 is a schematic diagram showing the four positions of the moving window, and Figure 3 is a schematic diagram showing the four positions of the moving window in the X-axis direction. These are characteristic diagrams for the X coordinate of the projection value, where (a) is a characteristic diagram corresponding to the window Wl in Figure 2, and (b) is a characteristic diagram corresponding to the window Wl in Figure 2.
is a characteristic diagram corresponding to window W2, (C) is a characteristic diagram corresponding to window W3, (d) is a characteristic diagram corresponding to window W4, and FIG. 4 is a characteristic diagram corresponding to window W4. These are characteristic diagrams for the Y coordinate of the projection value in the axial direction, where (a) is a characteristic diagram corresponding to window W1 in Figure 2, (b) is a characteristic diagram corresponding to window W2, and (C) is a characteristic diagram corresponding to window W2 in Figure 2. ) is the characteristic diagram corresponding to window W3,
FIG. 4(d) is a characteristic diagram corresponding to window W4. Code explanation DLD2: area, F: fixed image field of view, LLL2
: Straight line, P: Feature point, W, WLW2. W3. W4: Window. ■ ×○ Xz Muyo 5 people

Claims (1)

[Claims] 1) When two areas with different brightness are adjacent to each other with straight lines extending in two different directions from one point as a feature point as boundaries, the position of the feature point and the direction of each straight line. In the method of detecting a relationship between fixed images by image processing based on imaging, a rectangular window is generated in both fixed images including the feature point in the field of view, so that one of the straight lines passes through the window parallel to one side of the rectangular window. at each point when the window is moved parallel to the one straight line, a first line perpendicular to the direction of movement of the window;
taking a projection onto the side of and determining the coordinates of the starting point of change of the maximum value or minimum value of this projection relative to the first side of the window for each of the locations; based on the coordinates of each location; Detecting the position of the feature point; Taking a projection onto a second side perpendicular to the first side of the window for each of the locations, and determining its projection value; Increasing or decreasing the projection value for each location. A method for detecting feature point positions and straight line directions on a boundary line, comprising the following steps: detecting the directional relationship of each of the straight lines based on a tendency.