JPH0894541A - Method for detecting chipped corner of subject for inspection - Google Patents

Abstract

PURPOSE: To enable determining whether or not chipping has occurred regardless of the relation between the position of an-object for inspection and a circular window by processing image data for the subject for inspection, estimating the apex coordinates of a corner when the object is not chipped, setting the circular window whose center is located at the apex, and detecting whether or not the object is within the window. CONSTITUTION: An image of a subject 4 for inspection, sent from a camera 1, is input to an image processing device 2 and divided into the subject 4 and a background region. Equations for two straight lines sandwiching a corner forming the boundary between the subject 4 and the background region are calculated, and the coordinates of the crossing point of the straight lines are set as the apex of the corner. A circular window with a preset radius whose center is located at the crossing point is set. The radius is a value showing a criterion, and the smaller the value, the stricter the criterion. Whether or not chipping has occurred is determined according to whether or not a white image showing the presence of the subject 4 within the window is detected, and either an OK signal or an NG signal is output accordingly. This method eliminates the need for making corrections based on the relation between the position of the window and that of the subject.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an inspection method for automatically determining whether or not a corner of an inspection object having a corner such as a sheet has a minute chip.

[0002]

2. Description of the Related Art It is well known that a sheet, a plastic plate, a square or the like is likely to be chipped at its corner due to contact with other objects during the manufacturing process or during transportation. Recently, it has become possible to use robots for transportation,
For this purpose, it has become necessary to automatically detect the presence or absence of a chipped corner, which was conventionally visually inspected. As a method for detecting the lack of a corner, an inspection target is photographed by a camera or the like, and the captured image is subjected to image processing to calculate the entire area of the inspection target, and the calculated area and a preset reference are used. The presence or absence of a chip was determined from the difference between the area and the area. However, in this determination method, since the area of the entire inspection object is compared, there is a problem that it is not possible to detect a minute chip unless the accuracy of the image processing is high.

Therefore, in order to solve this problem, the vicinity of an arbitrary corner of the inspection object is photographed with a camera or the like, the photographed image is subjected to image processing, and a predetermined calculation standard is applied, so that there is no defect. A method of determining the presence or absence is used.
For example, in Japanese Unexamined Patent Publication No. 6-102203, the image output of a photographing device that photographs the corners of a building material is processed to detect a building material detection area and its background area, and an arbitrary plurality of areas on the boundary between these areas are detected. The equation of two straight lines sandwiching the corner is found from the coordinates of the points, and the virtual side that passes through the virtual intersection of the two straight lines obtained from this, that is, the point where the apex of the corner is assumed to be A method to automatically detect a chipping of a corner of a building material by setting a rectangular area (rectangular window) and determining whether the area ratio of the rectangular window and the detection area of the building material within this rectangular window is below a certain value. Is disclosed.
According to this method, since the designated area near an arbitrary corner of the inspection object is targeted for comparison, it is possible to detect even a minute defect with high accuracy without particularly improving the accuracy of image processing. It is possible.

[0004]

However, when setting a rectangular window from the virtual intersection obtained by this method, the size of the rectangular window, that is, the length of two sides of the rectangle of the window and the direction (angle) of the rectangular window. However, the area ratios of the rectangular window and the detection area of the inspection object in the rectangular window differ depending on the set values of these variables.
In particular, the area of the detection area of the inspection object in the rectangular window differs due to the difference in the orientation of the rectangular window and the positional relationship of the inspection object, so the area ratio also becomes completely different, and the reliability of the determination result decreases. Was causing the problem.

That is, when a rectangular window is used,
It is necessary to always keep the positional relationship between the direction of the rectangular window and the inspection object constant, but in the actual inspection process, for example, the process in which the inspection object sequentially flows on the conveyor, the orientation of the inspection object is constant. In most cases, the image processing device needs to take measures such as correcting the direction of the rectangular window in order to keep this positional relationship constant, which causes a problem that the processing time becomes long.
It should be noted that the above publication does not disclose nor suggest a method of correcting the positional relationship between the direction of the rectangular window and the inspection object.

Further, in the method disclosed in the above publication, the set values that influence the determination result are a total of four variables including the length and direction of the two sides of the rectangular window described above and the area ratio for determining the lack, There is also a problem that setting of judgment criteria becomes complicated.

The present invention has been made in order to solve the above-mentioned problems of the prior art, and the determination result is not affected by the positional relationship between the window and the inspection object, and moreover, the determination criterion which affects the determination result is set. It is an object of the present invention to provide a method for detecting a chipping of a corner of an inspection object that can simplify the above.

[0008]

Therefore, in the present invention, in a method for determining the presence or absence of a corner of an inspection object having a corner, first, an arbitrary inspection object having a corner is detected by a camera. The vicinity of the corner is captured, and the captured image is image-processed by the image processing device to be divided into the inspection object region in which the inspection object exists and the background region other than this region. Next, two straight lines that sandwich the corner portion that is the boundary between the inspection object region and the background region are extracted, and an arbitrary number of points are set on each of these two straight lines, and this arbitrary number is set. From the coordinates of the points, find the equation of two straight lines,
From the equation of the two straight lines, the coordinates of the virtual intersection, which is the intersection of the two straight lines, are obtained. Further, a circular window having a radius of a preset size centered on the virtual intersection is set in the captured image, and the inspection object is detected by detecting whether or not the inspection object area is included in the circular window. It was decided to solve the above-mentioned problems of the prior art by providing a method for detecting a corner defect of an inspection object, which is characterized by determining whether or not there is a corner defect.

Further, instead of detecting whether or not the inspection object area is included in the circular window, the area ratio between the circular window and the inspection object area in the circular window is calculated, and the calculated area ratio is calculated. The presence / absence of a corner of the inspection object may be determined by comparing area ratios that are preset determination criteria.

[0010]

According to the above-mentioned inspection method, the image data of the inspection object is processed, the coordinates of the apexes of the corners when the inspection object is not missing are estimated, and the estimated apexes of the corners are obtained. A circular window centered at is set, and the presence or absence of a defect is determined by whether or not the inspection object exists in this circular window. It becomes irrelevant to the position of the object.
In addition, the size of the circular window, that is, the radius, can be set as a criterion for determining the lack.

Further, instead of detecting whether or not the inspection object area is included in the circular window, the area ratio between the circular window and the inspection object area in the circular window is calculated. It is possible to determine the presence / absence of chipping of the corner of the inspection target by comparing the area ratio with the area ratio serving as a determination criterion set in advance. Also, since the area to be inspected is the area inside the circular window centered on the virtual intersection of the corners of the object to be inspected, the area of the area to be inspected in the window is irrelevant to the positional relationship between the window and the object to be inspected. Becomes

[0012]

Embodiments of the present invention will be described below with reference to FIGS. In particular, FIG. 4 is a flowchart showing a method of detecting a corner of the inspection object, and “(step **)” described in the text of this embodiment indicates a processing step in FIG. FIG. 1 is a schematic configuration diagram of an inspection apparatus to which an inspection method according to the present invention is applied. The inspection device is usually composed of a camera 1, an image processing device 2, and a monitor 3. An image of an inspection object 4 photographed by the camera 1 is input to the image processing device 2, and the image processing device 2 takes an image from the photographed image. The feature amount of the inspection object 4 is calculated using the processing means, and the result is displayed on the monitor 3. By the way, since the present invention is intended to determine the presence or absence of the chipped corners of the inspection object 4, the range of the input image is the inspection object 4 as shown in FIG.
It may be limited to the corners of the camera and its vicinity.
It may be carried out by adjusting the magnification of. Here, the broken lines at the corners shown in FIG. 2 indicate the case where there is no break, and the solid lines indicate the case where there are breaks.

The image of the inspection object 4 taken by the camera 1 is taken into the image processing device 2 (step 22) and divided into an area of the inspection object and its background area (step 23). The simplest and conventionally well-known division method is binarization. For example, when an inspection object having a bright surface is placed on a dark background, binarization causes the area of the inspection object to become white pixels, while the background area becomes black pixels. If the area where the pixel shifts from white to black is extracted, the boundary between the inspection object area and the background area is obtained (step 24). Another method is to use an edge enhancement filter or the like to extract only the contour line of the inspection object and separate the inspection object region and the background region. FIG. 3 shows an example of an image obtained by binarization. The shaded area is the inspection object area and is displayed white on the actual image processing screen, while the area other than the shaded area is the background area. Is displayed in black on the screen.

Next, in order to estimate the vertices of the corners of the inspection object, the equations of the two straight lines A and B sandwiching the corners that form the boundary between the inspection object area and the background area are obtained. (Step 25). The simplest method for obtaining the equation of a straight line is the known method of least squares. This method sets an X axis and a Y axis for an input image as shown in FIG. Several points are set at the points, and each linear function equation, that is, a linear equation is obtained from the coordinates of these points by using the least squares method.

Next, the coordinates of the intersections of the straight lines, that is, the vertices of the corners are obtained from the equation of the two straight lines that sandwich the corners (step 26). A method of obtaining the intersection point from the equations of two straight lines is also well known. When the equations of the straight lines A and B obtained above are expressed as shown in equations (1) and (2), respectively, the coordinates of the intersection points are expressed. (X1, Y1) is as shown in equation (3) and equation (4). The intersection of the obtained straight lines is estimated to be the apex of the corner of the inspection object, and this is called a virtual intersection.

[0016]

[Equation 1]

Further, as shown in FIG. 3, a circular window having a preset radius with the virtual intersection as the center is set (step 27). Here, the preset size is a numerical value used as a reference for determining the presence or absence of a chip, and if the numerical value is set small, the determination standard can be made strict. The presence / absence of a defect is determined by whether or not a white pixel indicating that the inspection object is present is detected in this circular window (step 2).
8) When there is an influence of noise or the like depending on the use environment, white pixels may be detected even though the inspection object does not actually exist in the circular window. Determining the presence or absence of a chip only by using it causes a problem in reliability. To deal with this, for example, if three or more adjacent pixels are white pixels in a row in a circular window, it is determined that a white pixel exists in the circular window. Good.

Finally, if the inspection object is detected in the circular window, it is determined that there is no defect, and an OK signal is output (step 29). On the other hand, if the inspection object is not detected in the circular window, the defect is detected. It is determined that there is, and an NG signal is output (step 30).

As described above, according to this embodiment,
The image data of the inspection object is processed, the coordinates of the vertices of the corners when the inspection object is not missing are estimated, and a circular window whose center is the estimated vertices of the corners is set, The presence / absence of a defect is determined depending on whether or not a white pixel indicating that the inspection object is present in the circular window is detected. In addition, the size of the circular window, that is, the radius, can be set as a criterion for determining the lack.

By the way, in the above-mentioned embodiment, the presence / absence of the defect is determined by whether or not the white pixel indicating that the inspection object is present in the circular window is detected, but as another means, By calculating the area ratio of the circular window and the area of the inspection object in the circular window, and comparing the calculated area ratio with the area ratio which is a preset judgment criterion, the presence or absence of the chip can be judged. Good. According to this method, the area ratio is added to the circular window in addition to the radius of the circular window, and the number of determination criteria increases compared to the above-described method of determining based on the presence or absence of white pixels. In a usage environment that is greatly affected by, the reliability of the determination result can be improved by this method that uses the area value that allows a certain degree of error as the comparison data. Further, in a process in which the inspection object sequentially flows on the conveyor, it is possible to prevent an abnormal state by performing process control by recording the area ratio as needed.

In the present embodiment, the case where the shape of the inspection object is a quadrangle is shown, but if the shape has corners, it may be another polygon such as a triangle, a hexagon, or an octagon. Alternatively, it may be fan-shaped or star-shaped, and the present embodiment can be applied.

[0022]

According to the present invention, the image data of the inspection object is processed to estimate the coordinates of the apexes of the corners when the inspection object has no defect, and the estimated apexes of the corners are estimated. A circular window centered at is set, and the presence or absence of a defect is determined by the inspection object detected in this circular window, so the determination result is irrelevant to the positional relationship between the window and the inspection object. . This eliminates the need for the process of correcting the position of the rectangular window in accordance with the position of the inspection object, which is necessary when using the rectangular window.

Further, since the size of the circular window, that is, the radius, can be set as the judgment criterion of the chip, the judgment criterion which influences the judgment result can be set simply and easily. In particular, in the method of determining the presence or absence of a defect by detecting whether or not a white pixel indicating that the inspection object exists in the circular window, the criterion for determination may be only the radius of the circular window. did it.

[Brief description of drawings]

FIG. 1 is a schematic diagram of an inspection system using image processing according to the present invention.

FIG. 2 is a diagram illustrating an example of an image input from a camera to an image processing device.

FIG. 3 is an explanatory diagram showing an embodiment of the present invention.

FIG. 4 is a flow chart showing a method of detecting a corner defect of an inspection object according to an embodiment of the present invention.

[Explanation of symbols]

 1 camera 2 image processing device 4 inspection object

Claims (2)

[Claims] 1. A method for determining the presence or absence of a corner of an inspection target having a corner, wherein a camera captures the vicinity of an arbitrary corner of the inspection target having the corner, and the captured image is obtained. Is subjected to image processing by an image processing device to divide into an inspection object region in which the inspection object exists and a background region other than the inspection object region, and a corner portion which is a boundary portion between the inspection object region and the background region. Extract the two straight lines that sandwich
Arbitrary several points are set on each of the two straight lines, the equations of the two straight lines are obtained from the coordinates of the arbitrary number of points, and the two straight lines are calculated from the equations of the two straight lines. The coordinates of the virtual intersection point that is the intersection point of the virtual intersection are determined, and a circular window having a radius of a preset size centered on the virtual intersection is set in the captured image, and the inspection object area is included in the circular window. The method according to claim 1, wherein the presence or absence of a corner of the inspection object is determined by detecting whether or not the corner of the inspection object is missing. 2. The area ratio between the circular window and the inspection object area in the circular window is calculated instead of detecting whether or not the inspection object area is included in the circular window, and the calculation is performed. The presence / absence of chipping of a corner of the inspection object is determined by comparing the determined area ratio with an area ratio serving as a preset determination standard.
A method for detecting a chipping of a corner of an inspection target described.