JP3675543B2 - Image position measurement method - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an image position measurement method, and more particularly, to measure the center position of an arbitrary circular object in an image in an image processing apparatus that performs position detection, measurement, inspection, and the like using an image signal as an input. The present invention relates to an image position measuring method.
[0002]
[Prior art]
For example, as a method for detecting the position of a circular object in an image captured by a CCD camera or the like, a binarization centroid method for detecting the position of the center of gravity after binarization processing or a pattern that matches a registered pattern A so-called pattern matching method has been used in the past.
[0003]
FIG. 6 is a diagram showing a state of position detection by the binarized centroid method. 6 shows the state of the captured image, the circular object, and the position of the center of gravity of the circular object obtained by the binarized center of gravity method, and FIG. 7 shows the state of position detection by the pattern matching method. In the pattern matching method, a pattern to be detected in advance is stored in the image processing apparatus as a registered pattern, and the same pattern as the stored pattern is searched from the captured image and output as the position of the detected pattern. These methods can detect the position of an image without any problem when only a circular object is captured in the captured image, but other than the circular object in the captured image. When a pattern is captured or when a part of a circular object in the captured image is missing, it is difficult to detect the pattern.
[0004]
[Problems to be solved by the invention]
FIG. 8 is a diagram illustrating an example of a case where a part of a circular object in the captured screen is missing. In the binarized centroid method, since the circular object is not binarized as a complete circle in the above-described case, the center position of the circular object and the centroid position may not match. Also, in the pattern matching method, the registered pattern and the pattern shape in the captured image do not match, so the degree of pattern matching decreases and the pattern position detection accuracy deteriorates.
[0005]
On the other hand, a feature quantity extraction method for extracting a feature from a captured image and detecting a circular object is also considered as a method for detecting the position of the circular object. According to the feature amount detection method, even when a pattern other than a circular target is captured in the captured image or when a part of the circular target in the captured image is missing, It is possible to detect the position.
[0006]
FIG. 9 is a diagram illustrating an example of the feature amount extraction method. In this example, the screen is sequentially searched radially and the position is estimated according to the conditions. As shown in the example, in the feature amount extraction method, the processing procedure is generally complicated, and it is necessary to perform operations on all the captured images, so that the amount of calculation is enormous.
[0007]
Therefore, even if the main object of the present invention is when a pattern other than a circular object is captured in the captured image, or when a part of the circular object in the captured image is missing. Another object of the present invention is to provide an image position measuring method capable of avoiding the problem that it becomes difficult to detect the center position of a circular object and preventing the calculation amount from becoming complicated and enormous.
[0008]
[Means for Solving the Problems]
The present invention, detection of the position of the image signal as an input, the measurement, an image processing apparatus for performing such testing, a plurality picked points Ru can become a point on the circumference of a circle of interest from the image represented by the image signal An image position measurement method that verifies whether or not any three of them are on the circumference of the target circle and calculates a candidate point for the center position of the target circle from the positions of the three points. Yes, in order to determine whether or not the three points are on the circumference of the target circle, the inclination of the tangent obtained for the circle estimated at the three points is compared with the inclination of the actual tangent. It is characterized by determining. Therefore , it is possible to perform processing at a relatively high speed with simple processing.
[0009]
Preferably, further, in order to three points to determine whether there is on the circumference of a circle of interest, it determines by comparing the radius of a circle whose radius and purpose of the circle to be estimated.
[0011]
Preferably, in order to extract correct center position data from a plurality of candidate point data calculated from a plurality of arbitrary three points in a plurality of sets, the data of the plurality of candidate points are classified as close to each other, and the most classified data Rights to smooth against frequent data group.
[0012]
DETAILED DESCRIPTION OF THE INVENTION
FIG. 1 is a schematic block diagram showing the configuration of an embodiment of the present invention. In FIG. 1, an image processing apparatus 1 stores a procedure for executing an image position detection method and sequentially executes the image position detection method. An image output apparatus 2 and an output apparatus 3 are connected to the image processing apparatus 1. Is done. The image output apparatus 2 includes an optical element, an imaging element, and peripheral circuits for inputting an image signal to the image processing apparatus 1. The output device 3 outputs the result of position detection processing by the image processing device 1. The image output apparatus 2 images the circular object 51 with an optical element or an image sensor. The circular object 51 exists on the workpiece 5 supported by the workpiece support 4. The circular object 51 is imaged by the image output device 2 and is arranged at a position where it is input to the image processing device 1 as an image signal.
[0013]
FIG. 2 is a flowchart showing a processing procedure for image position detection according to an embodiment of the present invention. The image output apparatus 2 shown in FIG. 1 captures an image of the circular object 51 and supplies it to the image processing apparatus 1, and the image processing apparatus 1 captures an image for one screen. Then, the image processing apparatus 1 extracts a so-called edge portion between the target object and other portions from the captured image. This processing may be performed while capturing the previous image. Further, since the following processing is also performed sequentially using the previous processing results, it can be executed by parallel processing if the mutual processing is not confused.
[0014]
Next, arbitrary three points are selected from the extracted edges, and candidate point extraction for calculating a center point of a circle predicted from the position information of the three points is executed. At this time, processing for speeding up the processing is performed, target point extraction is performed by collecting the central position data close to a plurality of extracted candidate points, and finally the target point position is obtained as a result. Output to the output device 3.
[0015]
FIG. 3 is a diagram for explaining a method of extracting a boundary between a pattern portion and other portions as an edge of the pattern in one embodiment of the present invention. Here, the edge position is estimated using multi-valued image data. The multi-valued image data will be described. First, the image data is stored in the image processing apparatus 1 as 512 discrete data in the horizontal direction and 480 discrete data in the vertical direction. Each piece of data is called a pixel. Each pixel is converted into numerical data in the range of 0 to 255 according to the brightness. That is, image data in which the brightness of the image at each position in the screen is multi-valued is obtained. In edge extraction, image coordinates that exceed a certain threshold value are calculated. Actually, in order to increase the accuracy of measurement, the edge coordinates are calculated by linearly complementing two image data sandwiching the threshold value. Note that the method of extracting an edge is not limited to this method.
[0016]
FIG. 4 is a diagram for explaining an edge extraction method. If the edge extraction is executed for all the screens, the amount of data handled becomes enormous. Therefore, the etching extraction is performed according to the edge scanning line as shown in FIG. That is, an edge scanning line is assumed along the horizontal and vertical directions and having a constant pitch interval, and edge extraction is performed on image data on the scanning line. The pitch of the edge scanning lines is set to an optimum value depending on the radius of the circular object 51 and the degree of chipping. Other methods may be used for the edge scanning method.
[0017]
Arbitrary three points are selected from the edge position information obtained by the edge extraction, and a circle (estimated circle) passing through the three points and its center (estimated center) are calculated. A circle passing through three points can usually be determined uniquely. At this time, the following two methods are used in order to distinguish an edge extracted from the circular object 51 from an edge extracted from other parts and prevent erroneous determination.
[0018]
(1) The radius of the circular object 51 to be measured is stored in advance and compared with the radius of the estimated circle passing through the three points.
[0019]
(2) Check the slope of the tangent at the edge point and verify whether it matches the circumferential direction.
[0020]
By these methods, processing time can be shortened and erroneous determination can be prevented. The estimated center obtained here is stored as a candidate point.
[0021]
First, the method (1) will be described. The center of the circle passing through the three points and its radius can be obtained by calculation. When the radius value obtained here is different from the radius of the circular object 51 to be measured, which is stored in advance, it is determined that the obtained estimated center data is an error.
[0022]
Next, the method (2) will be described. FIG. 5 is a diagram for explaining the effect of this tangential inspection method. Several edge extraction points are obtained by edge scanning as indicated by black circles in FIG. Consider an estimated circle that passes through the edge extraction points A, B, and C. Since the radius of the estimated circle is almost the same as that of the circular object 51, the method (1) cannot be excluded as an error. At this time, when the tangential slopes a, b, and c at the respective points are examined, the tangential slopes a ′, b ′, and c ′ at the respective points of the estimated circle are greatly different. In this way, by comparing the inclination of the tangent when assuming the point on the circumference of the circle estimated at each point and the inclination of the actual tangent, are the three points on the circumference? Can be determined.
[0023]
When a plurality of candidate points are obtained, adjacent candidate points are collected and the result is smoothed. The smoothing makes it possible to increase the position detection accuracy. Specifically, first, candidate points that are determined not to be adjacent are excluded from the processing, so that even if an erroneous candidate point is extracted by candidate point extraction, the data is excluded. Next, the data detected by the estimated positions are collected, and the number is calculated for each group. Finally, the data is averaged for one group having the highest frequency, and the position is set as the center position of the circular object 51. For smoothing, a method other than the method shown here may be used. And the obtained target point is output to the output device 3 as a result, and a series of processes are complete | finished.
[0024]
【The invention's effect】
As described above, according to the present invention, when the image position detection of a circular object by the binarized centroid method or the pattern matching method is performed, a part of the circular object is missing or a pattern other than the circular object exists. In some cases, the position of the image cannot be detected. In the present invention, candidate points that can be points on the circumference are selected from the image, and any three of them are the circumference of the target circle. Since the corresponding information is smoothed and output as the center position, it is possible to detect the position by specifying the edge extraction method, the radius of the circular object, etc. in advance It becomes possible. In addition, since the processing is simpler than the feature amount extraction method, processing can be performed at a relatively high speed.
[Brief description of the drawings]
FIG. 1 is a schematic block diagram showing the configuration of an embodiment of the present invention.
FIG. 2 is a flowchart for explaining the operation of the embodiment of the present invention.
FIG. 3 is a diagram for explaining a technique for extracting a boundary between a pattern portion and other portions as a pattern edge;
FIG. 4 is a diagram for explaining a method of extracting an edge according to an edge scanning line.
FIG. 5 is a diagram for explaining an effect of a connection inspection method.
FIG. 6 is a diagram illustrating a state of position detection by a binarized centroid method.
FIG. 7 is a diagram for explaining a state of position detection by a pattern matching method.
FIG. 8 is a diagram illustrating an example when a part of a circular object in a captured screen is missing.
FIG. 9 is a diagram illustrating an example of a feature amount extraction method.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 Image processing apparatus 2 Image output apparatus 3 Output apparatus 4 Work stand 5 Work 51 Circular target object

Claims (3)

In an image processing apparatus that performs position detection, measurement, inspection, etc. using an image signal as input,
Picked plurality of points from the image represented by the image signal Ru can become a point on the circumference of a circle of interest, whether any of the three points of which are on the circumference of a circle with the target This is an image position measurement method that verifies whether or not and calculates a candidate point of the center position of the target circle from the positions of the three points ,
In order to determine whether or not the three points are on the circumference of the target circle, the inclination of the tangent obtained for the circle estimated at the three points is compared with the inclination of the actual tangent. An image position measuring method characterized by determining . Further, characterized in that the three points in order to determine whether on the circumference of a circle with the target, determined by comparing the radius of a circle whose radius and the purpose of the circle to be estimated The image position measuring method according to claim 1. In order to extract correct center position data from the data of a plurality of candidate points calculated from a plurality of the arbitrary three points, the data of the plurality of candidate points is classified into adjacent ones, and the frequency of the most classified data 3. The image position measuring method according to claim 1, wherein smoothing is performed on a high data group.

Images