JP2001092959A - Method for setting and registering reference pattern and pattern matching recognizing method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method for setting and registering reference pattern, with which the position of a reference pattern can be exactly set and registered even when thinning processing is performed, and a pattern matching recognizing method for improving recognition accuracy. SOLUTION: Concerning this method for setting and registering reference pattern, a reference pattern 2 is displayed in the area of at least one part in the image of a pattern 1 to be detected, for which the image of an object is picked up, and a coarse recognition pattern 3 thinning the image data of this reference pattern 3 and a fine recognition pattern 4 highly accurately processing image data in the area of at least one part in the coarse recognition pattern 3 with this coarse recognition pattern 3 are displayed together and on the basis of both the displayed coarse and fine recognition patterns 3 and 4, the reference pattern 2 is set and registered on a characteristic part in the image of the pattern 1 to be detected.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a reference pattern setting registration method and a pattern matching recognition method.

[0002]

2. Description of the Related Art Conventionally, in a method of detecting the position of an electronic component and a board mark performed before mounting an electronic component by an electronic component mounting apparatus or the like, a reference pattern to be a reference is set and registered in advance, A pattern matching recognition method is used in which a reference pattern and a detected pattern to be recognized are compared by pattern matching to specify the position of the detected pattern.

This pattern matching recognition method is used in current electronic component mounting, and has a normal correlation coefficient as a method for recognizing the positions of various patterns to be detected such as leads of electronic components and board marks. Is used. Here, the pattern matching recognition method will be described. For example, consider image data 102 of a detected pattern 101 as shown in FIG. This is a teaching image for creating a reference pattern 103 used in pattern matching recognition. The reference pattern 103 shown in FIG. 4B is designated on the pattern 101 to be detected and shown in FIG. It is stored as image data inside the pattern window 104. (Teaching Step) Next, a method of detecting the position of the detected pattern 101 will be described.

In FIG. 5, when the detected pattern 101 exists inside the search area 105, the reference pattern 1
03 is sequentially moved into the search area 105 (Equation 1).
Is calculated for each coordinate. Among all the coordinates inside the search area 105, the coordinate at which the normal correlation coefficient has the maximum value is the position of the detected pattern 101 inside the search area 105. This (Equation 1) is an operation expression showing the operation content of the normal correlation coefficient.

Where γ (m, n) is a correlation coefficient at the coordinates (m, n) of the search area, R (i, j) is image data of the reference pattern, and S (m + i, n + j)
Is the image data of the search area, and a is the reference pattern size.

[0006]

(Equation 1) However, when the normal correlation coefficient is calculated by moving the reference pattern 103 for each pixel, there is a problem that the processing time increases. In order to solve this problem, a method of thinning out reference patterns is widely used.

A detected pattern 101 by thinning will be described below.
A method for detecting the position of the image will be described. Consider a coarse recognition pattern 106 as shown in FIG. The coarse recognition pattern 106 is obtained by omitting the pit data of the reference pattern 103 at even positions in the X and Y directions. The coarse recognition pattern 106 is sequentially moved inside the search area 105, and the normal correlation coefficient shown in (Equation 1) is calculated for each coordinate. Note that this is the same as calculating the normal correlation coefficient represented by (Equation 2). This (Equation 2) is an operation expression showing the contents of the operation of the normal correlation coefficient at the thinned pixel = 2. The coarse recognition coordinates 107 having the minimum value are detected.
In this calculation, the number of calculations is reduced to 1/4, and the processing time is accordingly reduced. However, using this calculated position alone, X
There will be a maximum of one pixel error in the Y direction. (Coarse Recognition Step) After that, there is a method of setting a fine recognition search area 108 corresponding to a thinning interval, and calculating a normal correlation coefficient for each coordinate based on the reference pattern 103 therein to detect the recognized coordinates. If the fine correlation pattern 109 composed of a part of the reference pattern 103 as shown in FIG. 7 is sequentially moved into the search area 108 to calculate the normal correlation coefficient, the processing time is further reduced. Become. (Fine Recognition Step) Here, L is reference pattern size / 2.

[0008]

(Equation 2) In the above-described conventional example, the thinning interval is set to one pit. However, when the thinning interval is set to N pits, the equation for calculating the normal correlation coefficient is as shown in (Equation 3). Is processed in the same way as This (Equation 3) is an operation expression showing the operation contents of the thinned pixel = N normal correlation coefficient.

Here, N is a thinned pixel + 1, and L is
Is the reference pattern size / N.

[0010]

(Equation 3)

[0011]

However, in the conventional pattern matching recognition method, when setting and registering the reference pattern 103 (teaching step), the operator monitors the coarse recognition pattern 106 with a rough image and monitors the characteristic of the detected pattern 101. This coarse recognition pattern 106 is set in a portion considered to be
The fine recognition pattern 109 is automatically displayed and set in the center portion of the reference pattern 103 instead of the coarse recognition pattern 106. The fine recognition pattern 109 which is a part of the reference pattern 103 is set.
However, if it is set to a flat part as shown in FIG. 8, the recognition position cannot be detected correctly, and even if the fine recognition pattern 109 can be automatically set to a characteristic part, this characteristic part is not accurate. If there is a problem in the detection position due to a problem such as the accuracy of the electronic component, such as a low portion, the coarse recognition pattern 106 is set again while monitoring the coarse recognition pattern 106 with a coarse image. It is necessary to repeatedly check the fine recognition pattern 109 and check whether the fine recognition pattern 109 is set at a high-accuracy portion in the characteristic portion of the pattern 101 to be detected. Pattern 10
There is a problem that it takes time to register the setting of No. 3.

If the reference pattern 103 is not set and registered in a characteristic part of the detected pattern 101 at a highly accurate location, an error occurs in the recognition of the detected pattern 101, and in the worst case, the electronic component There is a problem that the substrate receiving the mounting and the like is damaged. SUMMARY OF THE INVENTION An object of the present invention is to provide a reference pattern setting / registration method capable of accurately setting and registering the position of a reference pattern even when a thinning process is performed, and a pattern matching recognition method for improving recognition accuracy.

[0013]

According to a reference pattern setting / registering method of the present invention, a reference pattern is displayed in at least a partial area of an image of a detected pattern obtained by capturing an object, and image data of the reference pattern is displayed. The coarse recognition pattern and the fine recognition pattern obtained by processing the image data of at least a part of the coarse recognition pattern with higher precision than the coarse recognition pattern are displayed together. Based on the pattern, the reference pattern is set and registered in a characteristic part in the image of the detected pattern.

According to the present invention, it is possible to provide a reference pattern setting / registering method for setting and registering the position of a reference pattern accurately and efficiently in a short time even when a thinning process is performed.

[0015]

According to the first aspect of the present invention, a reference pattern is displayed in at least a part of an image of a detected pattern obtained by capturing an object, and image data of the reference pattern is displayed. A coarse recognition pattern that has been thinned,
The image data of at least a part of the coarse recognition pattern is displayed together with a fine recognition pattern obtained by processing the coarse recognition pattern with higher accuracy than the coarse recognition pattern. This is a reference pattern setting and registration method in which the reference pattern is set and registered in a characteristic portion of a pattern image. Even when thinning processing is performed, the position of the reference pattern is accurately and efficiently set and registered in a short time. A reference pattern setting registration method can be provided.

According to a second aspect of the present invention, there is provided a rough pattern obtained by displaying a reference pattern in at least a partial area of an image of a detected pattern obtained by capturing an object, and performing thinning processing on the image data of the reference pattern. A recognition pattern and a fine recognition pattern obtained by processing image data of at least a part of the coarse recognition pattern with higher accuracy than the coarse recognition pattern are displayed together, and based on the displayed coarse and fine recognition patterns. Setting and registering the reference pattern in a characteristic portion of the image of the detected pattern, and performing pattern matching comparison between the set and registered reference pattern and a detected pattern obtained by imaging an object to be recognized. This is a pattern matching recognition method that specifies the position of the detected pattern, and even if the thinning process is performed, the reference pattern Pattern matching that can be set and registered accurately and efficiently in a short period of time at a high-accuracy location in the characteristic portion of the detected pattern, reducing errors in recognizing the detected pattern and improving recognition accuracy A recognition method can be provided.

According to a third aspect of the present invention, there is provided a teaching step of registering a reference pattern which is a whole or a part of a pattern to be detected from an image of a detected pattern, and a camera image including the detected pattern. A coarse recognition step of detecting a coarse recognition coordinate having a maximum correlation coefficient with the search area of the coarse recognition pattern obtained by setting a search area and thinning out the registered reference pattern image data; and A fine recognition step of detecting a coordinate at which a correlation coefficient with a neighborhood area of the coordinate obtained in the coarse recognition step of the fine or full pattern of the pattern is a maximum, and In specifying the position, the rough recognition pattern area and the fine recognition pattern area are simultaneously displayed and the reference pattern are set and registered in the teaching step, A pattern matching recognition method for specifying the position of the detected pattern by comparing the recorded reference pattern and the detected pattern to be recognized with the coarse / fine recognition step in the pattern matching. Even when processing is performed, the position of the reference pattern can be set and registered accurately and in a short time with high accuracy in the characteristic part of the detected pattern, and errors in recognizing the detected pattern can be reduced. Thus, a pattern matching recognition method that improves recognition accuracy can be provided.

Hereinafter, a reference pattern setting registration method and a pattern matching recognition method according to the present invention will be described based on specific embodiments. (Embodiment) As shown in FIGS. 1 and 2, a pattern matching recognition apparatus according to an embodiment of the present invention includes a detected pattern 1 obtained by capturing an image of an object to be recognized and a reference whose setting is registered. A pattern matching recognizing device that specifies a position of the detected pattern 1 by performing pattern matching with a pattern 2, wherein a reference pattern 2 is placed on at least a part of an image of the detected pattern 1 obtained by capturing an object. A coarse recognition pattern 3 which is displayed and thinned out from the image data of the reference pattern 2;
And an image processing device 5 for displaying the image data of at least a part of the region together with the fine recognition pattern 4 which has been processed with higher precision than the coarse recognition pattern 3 and setting and registering the reference pattern 2 at the designated location. Things.

The reference pattern 2 is a pattern surrounded by a reference pattern window 2a, the coarse recognition pattern 3 is a pattern surrounded by a coarse recognition pattern window 3a, and the fine recognition pattern 4 is a fine recognition pattern. This is a pattern surrounded by a window 4a. Here, the operation of the pattern matching recognition device will be described.

As shown in FIG. 1, a detection target pattern 1 to be taught is arranged on a substrate 6. The detected pattern 1 represents an electronic component or a substrate pattern. An image input step of transferring image data taken from the camera 7 to the image processing device 5 is performed. FIG. 2 (a)
FIG. 4 is a front view showing an image to be taught. FIG.
In the image data 8 shown in FIG.
A search area setting step of setting a search area 9 which is an area to be inspected obtained from conditions such as the size of the search area is performed.

Next, a teaching step of setting and registering the reference pattern 2 is performed. An operator sets a reference pattern window 2a in a characteristic portion of the detected pattern 1 inside the search area 9. The reference pattern window 2a is composed of a coarse recognition window 3a and a fine recognition window 4a, and the coarse recognition window 3a and the fine recognition window 4a are simultaneously displayed on an image display unit (not shown) of the image processing device 5. The coarse recognition window 3a and the fine recognition window 4a move in synchronization. FIG. 2B is a front view showing the contents of the reference pattern window 2a. FIG. 3 shows the coarse recognition pattern 3 in the coarse recognition window 3a.
FIG. 7 is a front view showing the structure of internal data of a fine recognition pattern 4 in a fine recognition window 4a. The coarse recognition pattern 3 is obtained by omitting any thinned pit data in the X and Y directions inside the coarse recognition window 3a.
The fine recognition pattern 4 is data created without thinning out the data of the fine recognition window 4a.

As described above, the operator monitors the rough recognition window 3a and the fine recognition window 4a which are simultaneously displayed, and sets the reference pattern 2 to the detected pattern 1
Is set at a high-precision part in the characteristic part of and registered. After setting and registering the reference pattern 2 as described above, the same recognition process as in the above-described conventional example is performed. Specifically, the registered reference pattern 2 and the detected pattern 1 to be recognized are subjected to pattern matching in the coarse / fine recognition step described in the above-described conventional example, and the position of the detected pattern 1 is specified. I do.

With this configuration, even when a coarsely-recognized pattern subjected to the thinning process is used in the teaching process, the position of the reference pattern 2 can be accurately and quickly shifted to a high-accuracy portion in the characteristic portion of the detected pattern 1. Can efficiently register settings, reduce errors in recognizing the pattern 1 to be detected in the coarse / fine recognition process, improve recognition accuracy, and damage electronic components and the board on which they are mounted. The problem can be solved.

[0024]

As described above, according to the reference pattern setting / registering method of the present invention, the reference pattern is displayed in at least a part of the image of the detected pattern obtained by imaging the object, and the image of the reference pattern is displayed. A coarse recognition pattern obtained by thinning data and a fine recognition pattern obtained by processing image data of at least a part of the coarse recognition pattern with higher accuracy than the coarse recognition pattern are displayed together. By setting and registering the reference pattern in a characteristic portion of the image of the detected pattern based on the fine recognition pattern, the position of the reference pattern can be accurately and efficiently obtained in a short time even when the thinning process is performed. A reference pattern setting registration method for setting and registering can be provided.

According to the pattern matching recognition method of the present invention, a reference pattern is displayed in at least a part of an area of an image of a detected pattern obtained by capturing an object, and the image data of the reference pattern is thinned out. A coarse recognition pattern and a fine recognition pattern obtained by processing image data of at least a part of the coarse recognition pattern with higher accuracy than the coarse recognition pattern are displayed together, and based on the displayed coarse and fine recognition patterns. Setting and registering the reference pattern in a characteristic portion of the image of the detected pattern, and comparing the set and registered reference pattern with a detected pattern obtained by capturing an object to be recognized by pattern matching. Then, by specifying the position of the detected pattern, the position of the reference pattern can be accurately set even when the thinning process is performed. Registration can, it is possible to reduce the error in recognizing the detected pattern, the pattern matching recognition method to improve the recognition accuracy can be provided.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a pattern matching recognition device according to an embodiment of the present invention.

FIG. 2 is an exemplary front view showing an image to be taught according to the embodiment;

FIG. 3 is an exemplary front view showing a coarse recognition pattern and a fine recognition pattern according to the embodiment;

FIG. 4 is a front view showing a conventional image to be recognized.

FIG. 5 is a front view showing a conventional detected pattern and a reference pattern.

FIG. 6 is a front view showing a conventional coarse recognition pattern.

FIG. 7 is a front view showing a conventional fine recognition pattern.

FIG. 8 is a front view showing a conventional fine recognition pattern.

[Explanation of symbols]

 Reference Signs List 1 detected pattern 2 reference pattern 2a reference pattern window 3 coarse recognition pattern 3a coarse recognition window 4 fine recognition pattern 4a fine recognition window 5 image processing device 6 substrate 7 camera 8 image data 9 search area

Claims (3)

[Claims] 1. A rough recognition pattern in which a reference pattern is displayed in at least a part of an image of a detected pattern obtained by imaging an object, and a coarse recognition pattern obtained by thinning out image data of the reference pattern; A fine recognition pattern obtained by processing image data of a part of the region with higher precision than the coarse recognition pattern is displayed together, and based on the displayed coarse and fine recognition patterns, the image of the detected pattern is displayed. A reference pattern setting registration method for setting and registering the reference pattern in a characteristic portion. 2. A coarse recognition pattern in which a reference pattern is displayed in at least a part of an area of an image of a detected pattern obtained by capturing an object, and a coarse recognition pattern obtained by thinning out image data of the reference pattern; A fine recognition pattern obtained by processing image data of a part of the region with higher precision than the coarse recognition pattern is displayed together, and based on the displayed coarse and fine recognition patterns, the image of the detected pattern is displayed. The reference pattern is set and registered in a characteristic portion, and the position of the detected pattern is specified by performing pattern matching comparison between the set and registered reference pattern and a detected pattern obtained by imaging an object to be recognized. Pattern matching recognition method. 3. A teaching step of registering a reference pattern which is a whole or a part of a pattern to be detected from an image of a pattern to be detected, and a search area is set by registering a search area in a camera image including the pattern to be detected. A coarse recognition step of detecting a coarse recognition coordinate at which a correlation coefficient of the coarse recognition pattern obtained by thinning out the image data of the reference pattern with the search area is maximized; and a fine recognition that is a whole or a part of the registered reference pattern. A fine recognizing step of detecting a coordinate at which a correlation coefficient between the pattern and the vicinity area of the coordinate obtained in the coarse recognizing step is the maximum, and in specifying the position of the detected pattern, The area of the coarse recognition pattern and the area of the fine recognition pattern are simultaneously displayed, and the reference pattern is set and registered. Wherein the object detection pattern roughness, pattern matching recognition method and pattern matching compared with the seminal recognition process identifies the position of the detection target pattern to be.