JPH04323503A - Image processor - Google Patents

Abstract

PURPOSE:To calculate an accurate center coordinate by photographing a material having a circular contour, and by choosing only an appropriate data from image information. CONSTITUTION:A coordinate correspondent to the contour of an image data obtained by a photographing device A is extracted by a contour coordinate extraction device B, and a three-point coordinate is extracted from the extracted coordinates correspondent to the contour by a three-point extraction device C. The radius of a circle having the three-point coordinate on the circumference is calculated by a radius calculator E. According to the calculated radius, the appropriateness of the radius of the contour for determining a center coordinate is judged by a radius appropriateness comparator F. While the extracted three points tour around the contour extracted by the contour extraction device B, the extraction of three points, the calculation of radius, and the judgement of the appropriateness of the calculated radius are repeated by a three- point shifting device D, and a center coordinate correspondent to the three-point which is judged as appropriate by the radius fitting comparator F is calculated by center coordinate calculator G. After it is verified that the extracted three points are located on the same radius as that of a circle for calculation, the three points are used for calculation process of the center coordinate, and an accurate center coordinate is thus calculated.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus for capturing an image of an object having a circular contour and processing the image to determine the center coordinates of the circular contour. The circular contour here includes not only the outer contour of the workpiece but also the inner contour, such as the contour of a screw hole formed in the workpiece.

0002

[Prior Art] When the external outline of an object is circular, or when a circular outline such as a screw hole is formed on the surface of the object, this is imaged and image-processed to determine the center coordinates of the circular outline. A device has been developed to determine the This device determines the coordinates of the center of the object or the center of a screw hole, etc., and by instructing the robot to do this, the robot can work on the object, the screw hole, etc. This device extracts the coordinates corresponding to the contours of the external shape, screw holes, etc. from the imaged light-dark pattern, averages the extracted coordinate values, calculates the barycenter coordinates of the circular contour, and uses this as the center coordinates of the circular contour. Give instructions to robots, etc.

0003

[Problem to be Solved by the Invention] In order to accurately calculate the coordinates of the center of a circular contour of a screw hole, etc. using the above-mentioned conventional device, it is necessary to
The contour extracted as shown in FIG. 3(a) must be a perfect circle corresponding to the actual contour. However, if the inside of the screw hole is bright or a shadow is created around the screw hole due to the influence of lighting during imaging, and the difference in brightness between the screw hole and its surroundings is not necessarily clear, the extracted outline will be The shape is no longer a perfect reflection of the actual contour shape, and often ends up with parts missing, as shown in Figure 3(b), or parts sticking out, as shown in Figure 3(c). For this reason, the conventional apparatus has a problem in that the calculated barycenter coordinates do not correspond to the center coordinates of the circular contour. An object of the present invention is to solve the above problems. That is, the present invention calculates accurate center coordinates by selecting only appropriate data when calculating the center coordinates of a circle by processing image data obtained by imaging.

In order to solve the above-mentioned problems, the present invention provides a center coordinate calculation device whose concept is schematically shown in FIG. An image processing device that is an image capturing device A that captures an image of the object; a contour coordinate extraction device B that extracts coordinates corresponding to the contour of the object from image data obtained by the image capturing device A; a three-point extraction device C that extracts the coordinates of three points from the contour-equivalent coordinates extracted by the device B; and a radius of a circle having the coordinates of the three points extracted by the three-point extraction device C on its circumference. a radius calculation device E that calculates the radius calculation device E; and a radius suitability comparison device F that determines the suitability of the radius calculated by the radius calculation device E by comparing it with a predetermined value.
and the new extraction three-point positions extracted by the three-point extraction device are shifted from the previous extraction three-point positions3.
While the point shift device D and the three extracted points shifted by the three-point shift device D go around the contour-equivalent coordinates extracted by the contour coordinate extraction device B, the radius suitability comparison device F determines that the three points are suitable. It is characterized by comprising a center coordinate calculation device G that calculates the center coordinates of a circle having three points on the circumference, averages the calculated center coordinates, and calculates the center coordinates of the circular contour. Created an image processing device.

[0005]

[Operation] According to this device, the coordinates corresponding to the contour are extracted from the image data obtained by the imaging device A by the contour coordinate extraction device B. Then, the three-point extraction device C extracts the coordinates of three points from the extracted contour-equivalent coordinates. The radius calculating device E calculates the radius of a circle having the coordinates of these three extracted points on its circumference. Then, the radius suitability comparison device F determines whether this calculated radius matches the radius of the contour whose center coordinates are to be determined. The extraction of the three points, calculation of the radius, and determination of the suitability of the calculated radius are repeated by the three-point shift device D while the three extraction points go around the contour extracted by the contour extraction device B.
During this time, the center coordinate calculation device G calculates the center coordinates of the three points determined to be suitable by the radius suitability comparison device F, and then averages them. For this reason, according to the present device, only three points existing on the same radius as the contour whose center coordinates are to be determined are selected, and only the center coordinates calculated from the coordinates of these three points are used for averaging processing. Therefore, only correct data is selected from the image information, and this data is averaged to calculate correct center coordinates.

[0006]

DESCRIPTION OF THE PREFERRED EMBODIMENTS A first embodiment of the present invention will be described below with reference to FIGS. 2 to 5. This embodiment is an apparatus for calculating the center coordinates of a screw hole H formed in a workpiece indicated by I in FIG. 2, and shows an example in which the radius of the screw hole H is known. The calculated center coordinates are sent to the robot 10, and the robot 10 screws a screw into this screw hole.

A CCD camera 1 for taking an image of the work I is disposed above the work I in which the screw hole H is formed, with its field of view facing downward. The visual screen of the CCD camera 1 is composed of a large number of pixels arranged vertically and horizontally, and the position of each pixel on the visual screen can be extracted as a coordinate point. Furthermore, charges are generated in each pixel at a level corresponding to the brightness or darkness of the light that the individual pixel receives. The camera 1 is connected to an interface 4 connected to a bus 3 in order to send the image information represented on the visual screen to the CPU 2 as density information of individual pixels. The CPU2 is ROM5
and RAM 6 are connected to the bus 3, by which image information from the camera 1 is processed and the center coordinates of the screw hole H are calculated.

The ROM 5 stores various programs 1 that are executed to process image information sent from the camera 1.
01 to 112 are stored. The RAM 6 is provided with various storage areas 201-208 necessary for executing the programs 101-112. Furthermore, an operation panel 8 is connected to the interface 7 connected to the bus 3, and setting values such as parameters that are changed and set as appropriate according to the image information to be image processed are inputted from the outside. It has become. Also, robot 10 is on bus 3.
A robot CPU 9 is connected thereto, and the center coordinates of the screw hole H obtained through image processing are output to the robot CPU 9 .

When the workpiece I is imaged by the camera 1, a difference in pixel density occurs due to the difference in brightness between the screw hole H and its surroundings.
An image of the screw hole H appears on the visual screen. FIG. 3(a) exemplifies a case where the imaging state is ideal and the screw hole is accurately imaged. In this case, the center coordinates are calculated using the conventional method, that is, by averaging the contour coordinates and finding the barycenter coordinates. However, if the difference in brightness between the screw hole H and its surroundings is not necessarily clear due to conditions such as illumination during imaging, the difference in density between the screw hole H and its surroundings will not be clearly distinguished, so the difference in density between the screw hole H and its surroundings will not be clear. As shown in FIG. 3(c), the image of the screw hole H becomes an image b in which a part of the circle is missing, and an image c in which a part of the circle protrudes. The image captured on the visual screen of camera 1 in this way does not necessarily represent the actual screw hole H.
The circular outline of the image is often not clearly visible, resulting in an incomplete image. FIG. 4 shows an image G of the screw hole appearing in an incomplete state and a circular outline of the actual screw hole H superimposed.

Now, the center position h of the screw hole H is
It can be calculated from any three points a1 to a3 on the circular contour. This can also be understood from the fact that the center point of a circle coincides with the intersection of the perpendicular bisectors of each side of the triangle inscribed in the circle. On the other hand, for example, in the outline of the image G of a screw hole, at three points b1 to b3 located outside the outline of the actual screw hole H, a circle g having these three points b1 to b3 on the circumference is actually Since both the radius and the center position are different from the outline of the screw hole H, the correct center position of the screw hole H cannot be calculated. Therefore, the following processing is performed to select points corresponding to the outline position of the actual screw hole H from the image obtained by capturing the image with the camera 1 and calculate the center position of the screw hole.

First, after the counter K in the counter storage area 207 is cleared to zero in step S1 in FIG. Move to S3. In step S3, the image information binarization program 101 is executed, and the density information of each pixel input in step S2 is compared with a threshold, and the density information of the pixel is divided into a group of density levels above the threshold and a group of density levels below the threshold. Binarization processing is performed to divide the data into two groups. The threshold value used as the standard for the binarization process is a density level value that is approximately midway between the density of the pixel group in the part that makes up the image of the screw hole and the density of the pixel group in the part other than the image of the screw hole. has been done. Therefore, by binarizing using a threshold value, the portion of the image of the screw hole is separated from the other portions. Then, since the pixel group in the image of the screw hole is binarized to the density level side below the threshold value, in step S4, the position of the pixel on the visual screen for the image processed to the density level side below the threshold value. The coordinates indicating the are stored in the binarized image data storage area 201.

Next, in step S5, step S2
It is determined whether or not the processing up to S4 has been performed for all pixels, and if there are unprocessed pixels, step S4 is performed.
5 is NO, and steps S2 to S4 are repeated for unprocessed pixels. When the processing up to step S4 is completed for all pixels, the determination in step S5 is Y.
ES, and the process moves to step S6.

[0013] In step S6, the continuity discrimination program 102 is executed, and for the pixels of the density level corresponding to the screw hole stored in the binarized image data storage area 201 in the process of step S4, the pixel indicating this screw hole is selected. It is determined whether or not they are continuous with each other. If it is an inner part of the image of a screw hole, pixels with density levels corresponding to the screw hole are continuous in both the x-axis direction and the y-axis direction, but if it is a pixel that shows the outline of the image of the screw hole, , since the density level is changing and it is not continuous in either the x-axis direction or the y-axis direction, the coordinates of the pixel corresponding to the outline part among the pixels representing the image of the screw hole are extracted. and stored in the contour coordinate storage area 202.

Note that instead of this processing, the density information of the image may be differentiated to extract a region where the density suddenly changes, and coordinates corresponding to the contour may be extracted from this. Note that by determining the continuity, the positional relationship between the coordinates showing the images of the screw holes becomes clear, so if multiple screw holes H are imaged, the contour equivalent of the image of each screw hole is determined. The coordinates are extracted separately from the contour-equivalent coordinates of images of other screw holes.

Next, in step S7, the first parameter setting program 103, the second parameter setting program 104, and the third parameter (ΔR) setting program 10 are executed.
5, and the known radius (R) setting program 106 is executed, and the first parameter, second parameter, third parameter (ΔR), and known radius (R) are set by input operation from the operation panel 8, Each set value is stored in the first parameter storage area 203, the second parameter storage area 204, and the third parameter storage area 203.
The parameters are stored in the parameter (ΔR) storage area 205 and the known radius (R) storage area 206, and the process moves to step S8.

In step S8, the three-point extraction program 107 is executed, and the coordinates of three points are set in step S7 from among the coordinates corresponding to the contour of the screw hole stored in the contour coordinate storage area 202 in the process of step S6. Extracted based on the first parameter and the second parameter. Here the first
The parameter determines the extraction interval between the coordinates of three points, and the second parameter shifts the extraction position relative to the previous extraction position when extracting the coordinates of another three points based on the first parameter. This determines the amount of For example, in Fig. 4, if the coordinates of three points m1, m2, m3 are extracted, the coordinates between m1 - m2 and m2 - m3
Each extraction interval in between is determined by a first parameter P1. Then, the coordinates of the next three points extracted n1, n
2, n3, the coordinate n1 is shifted to the position determined by the second parameter P2 with respect to the previously extracted coordinate m1, and the coordinates n1 and n2 are shifted to the position determined by the second parameter P2.
Each extraction interval in between is determined by a first parameter P1. In this way, the coordinates of three other points are extracted one after another from among the contour-equivalent coordinates.

After the coordinates of the three points are extracted in step S8, it is determined in step S9 whether or not the extraction positions of the three points have gone around the outline of the image of the screw hole from the first extraction position. If not, the determination is YES and the process moves to step S10. In step S10, the radius calculation program 108 and the center calculation program 109 are executed, and the radius Ri of the circle having the coordinates of the three points extracted in step S8 on the circumference and the center coordinates (Xi, Yi) of the circle are calculated. Each is calculated. That is, as described above, the center coordinates (Xi, Yi) of the circle are calculated from the three points extracted in step S8, and the center coordinates (Xi, Yi) are calculated from the three points extracted in step S8.
) and one of the three points, the radius Ri is calculated.

Next, in step S11, the radius comparison program 110 is executed, and the size of the radius Ri calculated in step S10 is compared using the third parameter (ΔR) set in step S7 and the known radius (R). A judgment is made. Here, the known radius (R) is the radius of the screw hole H, and the third parameter (ΔR) is the allowable error value of the known radius (R).

As mentioned above, if all of the coordinates of the three points extracted from the contour-equivalent coordinates of the image of the screw hole are coordinates located on the actual contour of the screw hole H, the coordinates of the three points are When the size of the radius Ri of the circle having the coordinates on the circumference matches the radius R of the screw hole H, but the coordinates of the three points include coordinates of a position that is outside the actual outline of the screw hole H. for,
Since the radius Ri does not match the radius R of the screw hole H, the actual screw hole H can be determined by determining whether the radius Ri is within the error range of the radius R of the actual screw hole H. Only the coordinates corresponding to the contour of can be selected.

The size of the calculated radius Ri is the radius R of the screw hole H.
If the coordinates of the three points extracted in step S8 are within the error range of , that is, if the coordinates of the three points extracted in step S8 are located on the outline of the actual screw hole H, the determination in step S11 is YES, and the counting program 11 is executed in step S12. , 1 is added to the counter K in the counter storage area 207, and then the process moves to step S13. In step S13, the center coordinate update program 112 is executed, and the center coordinates (Xi, Yi) calculated in step S10 are changed to the center coordinates (Xs,
Ys) is adopted for updating.

That is, the center coordinates (Xs, Ys) are updated by calculating the average value of the center coordinates (Xi, Yi) calculated in step S10 for all the cases where the determination in step S11 is YES. The newly averaged center coordinates (Xs, Ys) including the center coordinates (Xi, Yi) calculated in step S10 are stored in the center coordinate storage area 208. When the process of step S13 is completed, the process moves to step S8, and coordinates of three points different from the previous time are extracted from among the contour-equivalent coordinates of the image of the screw hole.

[0022] In step S11, if the size of the radius Ri is not within the error range of the radius R of the screw hole H, that is, if the coordinates of the three points include coordinates of a position outside the actual contour of the screw hole H. The determination is NO and the process moves to step S8, and the center coordinates (Xi, Yi) calculated in the process of step S10 this time are the center coordinates (X
s, Ys) is not adopted for updating.

When the extraction of the coordinates of the three points is repeated in step S8 and the extraction position is shifted from the first extraction position to a position that goes around the outline of the image of the screw hole, the determination in step S9 becomes NO. Center coordinates (Xs, Ys) stored in the center coordinate storage area 208 in step S14
is the center position (X, Y) of the screw hole H, and the robot CP
It is output to U9 and all processing ends. Then, the robot C based on the detected center position (X, Y) of the screw hole
The operation of the robot 10 is controlled by commands from the PU 9, and a screw is attached to the screw hole H of the workpiece I.

According to this example, since the center position of the screw hole H can be calculated by selecting the coordinates corresponding to the outline of the actual screw hole H from the image taken by the camera 1, the illumination at the time of image sensing can be calculated. Even if the image of the screw hole H cannot necessarily be obtained in a perfect state due to influences or the like, the center position of the screw hole H can be accurately detected. Also, in this example,
In the case where there are two holes, for example, when there is an actual screw hole at the bottom of the screw through hole, the center coordinates calculated from the contour coordinates of the screw through hole are calculated in step S.
Only the center coordinates calculated from the actual outline coordinates of the screw hole are extracted and the center coordinates of the screw hole are calculated. Therefore, in the conventional processing method, accurate center coordinates are calculated even when there is a high possibility of erroneous calculation.

Next, regarding the second embodiment of the present invention, FIGS.
With reference to FIG. 8, descriptions of parts similar to those in the first embodiment will be omitted, and parts that are different will be described. The device of this example is capable of calculating the center coordinates of a screw hole whose size is unknown. Program 1 similar to the first embodiment is stored in ROM5A.
In addition to 01A to 108A, programs 301 to 305 are executed for processing when the size of the screw hole H is unknown.
are stored in the RAM 6A, in addition to the storage areas 201A to 204A similar to those in the first embodiment, the programs 301 to 3 are stored in the RAM 6A.
Storage areas 401 to 403 necessary for executing 05 are provided.

Through the processing up to step S10A (see FIG. 7) similar to the first embodiment, the radius Ri and After the center coordinates (Xi, Yi) are calculated, the process moves to step S101. However, in step S7 mentioned above, the third parameter (ΔR), the known radius (R
) is set, whereas in step S7A, the lowest frequency (T), which will be described later, is set. Step S101
In , the radius classification program 301 is executed, and based on the size of the radius Ri calculated in step S10A,
Center coordinates (Xi, Yi) calculated in step S10A
A classification is performed. Here, the classification divisions provided for the size of the radius for each ΔR...Rj-1, Rj, Rj
+1..., the center coordinates (Xi, Yi) are classified for each division Rj to which the radius Ri corresponds. FIG. 8 shows an example of classification based on the radius Ri obtained by the above processing.

Next, in step S102, the counter program 302 is executed, and the counter storage area 4 for each radius classification is
After 1 is added to the counter of the corresponding classification section in the process of step S101 among the counters provided corresponding to each classification section. In step S103, the center coordinate update program 303 is executed, and for the center coordinates (Xj, Yj) of the corresponding section in the process of step S101, the average value (Xj ,Yj
) and the center coordinates (Xi, Yi) calculated in the process of step S10A this time are averaged and updated,
The newly calculated center coordinates (Xj, Yj) are stored in the center coordinate storage area 402 for each radius classification separately from the center coordinates for other classification sections. When the process of step S103 is completed, the process moves to step S8A, the coordinates of another three points are extracted, and steps S9A to S103 are repeated.

[0028] The extraction of three points in step S8A is repeated until the outline of the image of the screw hole is circled once, and if the determination is NO in step S9A, the process moves to step S104, and the counter storage area 401 for each radius classification is After the total count number K, which is the sum of the count numbers k0, k1 . . . kn factored into each counter, is calculated, the process moves to step S105. In step S105, the radius frequency comparison program 304 is executed, and the count number kj of each counter is compared and determined with the lowest count frequency K.T. Here, the lowest frequency (T) is set by executing the lowest frequency setting program 305 in step S7A as described above, and is stored in the lowest frequency storage area 403, and T
A value of approximately 0.8 to 0.9 is adopted.

Regarding the radius Ri of the circle determined by the coordinates of three points extracted from the contour-equivalent coordinates of the image of the screw hole,
When the radius Ri of a predetermined size is calculated significantly more,
That is, when the contour-equivalent coordinates of the screw hole image are concentrated on the circumference of a predetermined size, it can be determined that coordinates indicating the actual contour position of the screw hole H exist on the circumference. This can be determined regardless of whether the size of the screw hole H is known or unknown, so in this example, the actual contour position of the screw hole H is shown from the contour equivalent coordinates of the screw hole image. In order to extract the coordinates, the count number kj of each division classified by the size of the radius Ri, that is, the radius R of the size of each division
It is determined whether the number of calculations of j is greater than the total count number K, that is, the minimum frequency (T) allowable for the total number of calculations.

Therefore, unless the count number kj of the counter has reached the minimum frequency K·T of the count number, the coordinates indicating the actual contour position of the screw hole H cannot be extracted, so the determination in step S105 is NO, and the step After the counter is changed to another classification category in S106, the counter in step S10
Returning to step 5, the count number kj is compared with the lowest frequency K.T. Then, the compared count number kj is the lowest frequency K
- If T is exceeded, the coordinates indicating the actual contour position of the screw hole H can be extracted from that division, so the determination in step S105 is YES, and the center coordinates (Xj, Yj) in step S10
At step 7, the center coordinates (X, Y) of the screw hole H are determined, and all processing ends.

According to this example, even if the size of the screw hole H is unknown, the coordinates corresponding to the actual contour position of the screw hole H are extracted from among the coordinates corresponding to the contour of the image of the screw hole, and the screw hole is determined. The center position of H can be calculated accurately. In the first embodiment, since the radius of the circle whose center coordinates are to be determined is known, the radius suitability comparison device F uses the known radius as is. On the other hand, in the second embodiment, since the radius of the circle is unknown, it is not possible to determine whether the correct three points have been extracted before processing. Therefore, in this embodiment, the extracted radius that appears more frequently than the minimum frequency is set as the appropriate radius, and the radius suitability comparison device F uses this as a reference value for comparison. In any of the embodiments, the three extracted points are used to calculate the center coordinates after it is confirmed that they are located on the same radius as the circle to be calculated, so processing based on incorrect data may occur. is not performed, and accurate center coordinates are calculated.

[0032]

[Effects of the Invention] According to the present invention, even if an image in a perfect state is not necessarily obtained, appropriate coordinates located on the circumference of the circle whose center is to be found from image data are selected, and the center coordinates of the circle are determined. can be calculated accurately, so it is particularly effective for objects for which it is difficult to set appropriate imaging conditions and for which it is difficult to obtain a clear image.

[Brief explanation of drawings]

FIG. 1 is a diagram schematically showing the concept of the present invention.

FIG. 2 is a diagram showing the overall configuration of the device according to the first embodiment of the present invention.

FIG. 3 is a diagram showing an example of an image of a screw hole.

FIG. 4 is a diagram showing a screw hole and an image of the screw hole in a superimposed state.

FIG. 5 is a flowchart showing processing in the apparatus according to the first embodiment of the present invention.

FIG. 6 is a diagram showing the overall configuration of an apparatus according to a second embodiment of the present invention.

FIG. 7 is a flowchart showing processing in an apparatus according to a second embodiment of the present invention.

FIG. 8 is a histogram showing an example of processing in the apparatus according to the second embodiment of the present invention.

[Explanation of symbols]

1,1A camera 2,2A CPU 5,5A ROM 6,6A RAM H Screw hole

Claims (1)

[Claims] 1. An image processing device that images an object having a circular contour and performs image processing to calculate center coordinates of the circular contour, comprising: an imaging device that images the object; a contour coordinate extraction device that extracts coordinates corresponding to the contour of the object from image data; a three-point extraction device that extracts coordinates of three points from among the contour-equivalent coordinates extracted by the contour coordinate extraction device; a radius calculation device that calculates the radius of a circle having the coordinates of the three points extracted by the three-point extraction device on the circumference; and a radius calculation device that determines whether or not the radius calculated by the radius calculation device is appropriate by comparing it with a predetermined value. The radius suitability comparison device and the three-point extraction device extract new three-point positions to positions shifted from the previous three-point extraction positions3.
While the point shift device and the three extraction points shifted by the three-point shift device go around the contour-equivalent coordinates extracted by the contour coordinate extraction device, the three points determined to be suitable by the radius suitability comparison device are An image processing device comprising: a center coordinate calculation device that calculates center coordinates of a circle on the circumference, averages the calculated center coordinates, and calculates center coordinates of the circular contour.