JP2009068998A - Position-detecting device and position-detecting method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a position-detecting device capable of improving matching accuracy by eliminating pseudo-matching, and capable of making the template matching processing time improved. <P>SOLUTION: The device for detecting the position of a mark pattern that coincides with a template in a retrieving target image is provided with a retrieving target image processing part 20 for performing image processing of the retrieving target image so as to extract a part that varies between adjacent pixels, in pixels that constitute the image; and a detecting part (template matching processing part 40) for detecting a position of the mark pattern coinciding a template image of the template in the retrieving target image, to which the image processing is applied. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a position detection apparatus and a position detection method for detecting the position of a mark pattern that matches a template in a search target image by image processing.

  In an apparatus for detecting a position using an image, an object to be detected is registered as a template in advance, template matching between the template image of this template and image data to be searched is performed, and a search target image that matches the template image The position of the mark pattern inside is detected (Patent Document 1).

  However, depending on the state of the search target itself, for example, if the search target is a semiconductor wafer, the state of the surface of the semiconductor wafer, or the condition (contrast and noise) of the input image obtained by photographing the search target, A pseudo pattern similar to the mark pattern to be detected may occur.

  In order to avoid such a pseudo matching that matches the pseudo pattern, it is necessary to discriminate the pseudo pattern from the mark pattern on the search target image and finally find the mark pattern.

Also, if there is a possibility that the gradation of the image of the mark pattern is reversed depending on the state of the search target image, two types of templates are prepared in advance, and one of the two types of templates is used. It is necessary to perform template matching for two times so that template matching is performed and then template matching is performed using the other template to find the position of the matched mark pattern.
JP 2000-155011 A

  However, the task of discriminating between the pseudo pattern and the mark pattern on the search target image requires skill, cannot avoid the possibility of the occurrence of pseudo matching that matches the pseudo pattern, and there is a problem that matching accuracy decreases. there were.

  In addition, in order to allow a mark pattern to be inverted while searching for a mark pattern, it is necessary to perform template matching for two times, which increases the processing time for template matching.

  The present invention has been made in view of the above circumstances, and provides a position detection device and a position detection method capable of improving pseudo-matching to improve matching accuracy and shortening template matching processing time. Objective.

  The position detection device according to claim 1 of the present invention that achieves the above object is a device that detects the position of a mark pattern that matches a template in a search target image, and the image is configured in the search target image. A search target image processing unit that performs image processing to extract a portion that changes between adjacent pixels among the pixels to be matched with a template image of the template in the search target image that has been subjected to the image processing And a detection unit for detecting the position of the mark pattern.

  The position detection apparatus according to claim 2 of the present invention includes a template image processing unit that performs image processing for extracting a part that changes between adjacent pixels from the pixels constituting the image in the template image. The mark that matches the template image that has been subjected to the image processing by the template image processing unit in the search target image that has been subjected to the image processing by the search target image processing unit The position of the pattern is detected.

  The position detection apparatus according to a third aspect of the present invention is characterized in that the image processing is processing for performing edge enhancement filter processing in two directions intersecting each other.

  According to a fourth aspect of the present invention, the mark pattern is composed of a plurality of element patterns, and the element patterns are smaller than a processing area of the edge enhancement filter process.

  The position detection apparatus according to claim 5 of the present invention is characterized in that the element pattern is an image having a corresponding number of pixels of 2 × 2.

  In the position detection device according to a sixth aspect of the present invention, the edge enhancement filter process performs the edge enhancement filter process on the image in one of the two directions, and then the edge enhancement filter in the one direction. The edge enhancement filter process is further performed in the other direction on the processed image.

  According to a seventh aspect of the present invention, in the position detection device, the edge enhancement filter process performs the edge enhancement filter process on the image in one direction out of the two directions, and the edge enhancement filter process on the image in the other direction. And the two images subjected to the edge enhancement filter processing are overlapped, and the overlapped portion is defined as the image.

  A position detection method according to an eighth aspect of the present invention is a method for detecting the position of a mark pattern that matches a template in a search target image, and the template image of the template is edged in two directions intersecting each other. In the template processing step for performing enhancement filter processing, the search target image processing step for applying edge enhancement filter processing to the search target image in two directions intersecting each other, and the search target image subjected to the image processing, And a step of detecting a position of the mark pattern that matches the template image subjected to the image processing.

  According to the present invention, pseudo matching can be reduced, matching accuracy can be improved, and template matching processing time can be shortened.

  Hereinafter, an embodiment of a position detection apparatus and a position detection method of the present invention will be described with reference to FIGS.

  In the position detection apparatus and position detection method of the present embodiment, a vertical edge enhancement filter process (hereinafter referred to as a vertical edge enhancement filter process) and a horizontal direction are applied to a template image (pattern image) and a search target image, respectively. Both image processing of edge enhancement filter processing (hereinafter referred to as horizontal edge enhancement filter processing) is performed, and template matching is performed using the image subjected to these filter processing.

  FIG. 1 is a schematic block diagram showing an embodiment of the position detection apparatus of the present invention.

  The position detection apparatus according to the present embodiment includes a template image processing unit 10 that performs vertical / horizontal edge enhancement filter processing on a template pattern image, and a search target image processing unit 20 that performs vertical / horizontal edge enhancement filter processing on a search target image. The store unit 30 that stores the filtered image in the template image processing unit 10 and the search target image processing unit 20, and the template matching between the pattern image of the filter processed image stored in the store unit 30 and the search target image is performed. A template matching processing unit 40 as a detection unit to be executed, and a control unit 50 that controls the processing units 10, 20, 40 and the store unit 30 are provided.

  As shown in FIG. 2, the template image processing unit 10 includes an image acquisition unit 11 that inputs a pattern image for a template, and a filter processing unit 12 that performs vertical / horizontal edge enhancement filter processing on the input image. Similarly, the search target image processing unit 20 includes an image acquisition unit 21 that inputs a search target image and a filter processing unit 22 that performs vertical / horizontal edge enhancement filter processing on the input image.

  By the vertical / horizontal edge enhancement filter processing, the pseudo patterns in the search target image are significantly reduced (see FIG. 12), and the matching accuracy is improved. As a result, the mark pattern can be reduced.

  Further, the vertical and horizontal edge emphasis filter processing causes the image with the inverted gradation of the image and the non-inverted image to have the same pattern, so that the pattern image is not affected by the inversion of the gradation of the mark pattern. There is no need to prepare a type, and the template matching process is only required once, and the processing time can be shortened.

  FIGS. 6A and 6B are explanatory diagrams of a vertical edge enhancement filter and a horizontal edge enhancement filter used for vertical / horizontal edge enhancement filter processing in the template image processing unit 10 and the search target image processing unit 20.

  When the convolution operation is executed on the input image (pattern image, search target image) using the kernel 60 (FIG. 6A) of the vertical edge enhancement filter, the horizontal edge component is eliminated from the input image, and the vertical direction Only the edge component remains. This is because the difference between adjacent pixels in the horizontal direction of the input image (the difference in gradation from the adjacent pixels) is used as the pixel value of the output image.

  Similarly, when the convolution operation is executed using the kernel 61 (FIG. 6B) of the horizontal edge enhancement filter, the vertical edge component disappears from the input image, and only the horizontal edge component remains. This is because the difference between adjacent pixels in the vertical direction of the input image is used as the pixel value of the output image.

  Since both the processing using the vertical edge enhancement filter and the processing using the horizontal edge enhancement filter output the difference from adjacent pixels, the same image can be obtained even if the gradation is inverted. Thus, even if there is a possibility that the gradation is inverted, template matching can be performed without preparing two types of templates.

  For example, when the gradation of one adjacent pixel is 255 gradation and the gradation of the other pixel is 0 gradation, the difference 255 (255-0) between the two is the pixel value of the output image. The gradation of this adjacent pixel is inverted, the gradation of one adjacent pixel changes from 255 gradation to 0 gradation, and the gradation of the other pixel changes from 0 gradation to 255 gradation. However, the difference between them is 255 gradations (255-0), and this is the pixel value of the output image, so the same image can be obtained. When the gradations of adjacent pixels are both 255 gradations or 0 gradations, the difference between them is 0 gradations, and the pixel value of the output image is 0 (no output image).

  In other words, the part that changes between adjacent pixels is extracted by the filtering process, and the other part that does not change is removed.

  FIG. 3 is a flowchart showing an embodiment of the position detection method of the present invention.

  In step S1, the template pattern image is processed by combining vertical edge enhancement filter processing and horizontal edge enhancement filter processing, and converted to a template image used in the present embodiment.

  When template matching is performed using the same template, this step S1 may be executed once, and step S1 need not be executed for each template matching. When the same template is not used, step S1 is executed each time.

  In step S2, the search target image is processed by combining the vertical edge enhancement filter processing and the horizontal edge enhancement filter processing, and is converted into a search target image used in the present embodiment.

  In step S3, processing for obtaining the position of the mark pattern by template matching is executed using the images obtained in steps S1 and S2.

  FIG. 4 is a first flowchart specifically showing the contents of the combination of the vertical edge enhancement filter process and the horizontal edge enhancement filter process executed in steps S1 and S2 of FIG.

  As shown in FIG. 4, in step S4, vertical edge enhancement filter processing is executed on the input image (template pattern image, search target image), and then in step S5, the image after vertical edge enhancement filter processing is executed. Further, horizontal edge enhancement filter processing is executed.

  In step S4, horizontal edge enhancement filter processing is executed instead of vertical edge enhancement filter processing, and then vertical edge enhancement filter processing is further executed on the image after horizontal edge enhancement filter processing in step S5. Good.

  FIGS. 7 and 8 show the result of executing the filter process by the method shown in FIG. 4 on the mark pattern in the search target image.

  FIG. 7 shows a result of executing the vertical edge enhancement filter process and then executing the horizontal edge enhancement filter process on the image after the vertical edge enhancement filter process. FIG. 5A shows a mark pattern 70 in the search target image used in the present embodiment. In this mark pattern 70, an element pattern (hereinafter referred to as an element) 72 formed by combining four (2 × 2) pixels 71 has an interval corresponding to two pixels 71 in the vertical and horizontal directions. A plurality of them are arranged. The dimension of the mark pattern 70 is set so as to maintain the arrangement relationship between the elements 72 (interval between the elements 72). For example, in this embodiment, when the size of the pixel 71 is 1 mm square, if the size of the element 72 is 2 mm square and the interval between the elements 72 is 2 mm, the mark pattern 70 has a size of 10 mm square.

  When this mark pattern 70 is processed by the vertical edge emphasis filter, the mark pattern 70 shown in FIG. 7A is changed to a mark pattern 74 having a plurality of horizontally long elements 73 as shown in FIG. 7B.

  In FIG. 7A, for example, when the gradation processing is performed by setting the gradation of the pixel 71 in the hatched portion with a hatched pattern to 255 gradation and setting the pixel 71 in the white background portion without the hatched pattern to 0 gradation. In the pixel 71a and 71b adjacent to each other in the horizontal direction, the gradation of the pixel 71a is 0 gradation, the gradation of the adjacent pixel 71b is 255 gradation, and the difference between them is 255 gradation. This is the pixel value of the output image. Similarly, regarding the pixels 71c and 71d adjacent to each other, the pixel 71c has 255 gradations, the pixel 71d has 0 gradation, and the difference between them is 255 gradations, which is the pixel value of the output image. On the other hand, regarding the pixel 71e and the pixel 71f, both the pixel 71e and the pixel 71f have 0 gradation, and the difference between them is 0 gradation, which is the pixel value of the output image.

  That is, in the horizontal direction (X-axis direction) of the mark pattern 70, the gradation of the pixel 71 between the elements 72 becomes 255 gradation, and changes to the mark pattern 74 having the horizontally long element 73 as described above.

  Next, if the mark pattern 74 after this vertical edge enhancement filter processing is further processed by the horizontal edge enhancement filter, the gradation of the pixel 71 between the elements 73 in FIG. Changes from 0 gradation to 255 gradation, and changes to a mark pattern 76 having a rectangular element 75 as shown in FIG.

  In FIG. 7B, for example, regarding the pixels 71a and 71h adjacent to each other in the vertical direction (Y-axis direction) of the mark pattern 76, the gradation of one pixel 71a is 255 gradations, and the other pixel 71h The gradation is 0 gradation, and the difference between the two is 255 gradation, which is the pixel value of the output image. Similarly, regarding the pixels 71g and 71e adjacent to each other, one pixel 71g has 255 gradations, the other pixel 71e has 0 gradation, and the difference between them is 255 gradations. Value.

  As described above, when the filtering process is executed by the method shown in FIG. 4, the mark turn 76 of the image shown in FIG. 7C is searched by the pattern matching executed in step S3 of FIG.

  FIG. 8 shows the result of executing the horizontal edge enhancement filter process and then performing the vertical edge enhancement filter process on the image after the horizontal edge enhancement filter process, unlike the case shown in FIG. FIG. 7A shows a mark pattern 70 in the search target image used in this embodiment, which is the same as that shown in FIG.

  When this mark pattern 70 is processed by the horizontal edge enhancement filter, the mark pattern 70 shown in FIG. 8A is changed to a mark pattern 78 having a plurality of vertically long elements 77 as shown in FIG. 8B.

  That is, in the vertical direction of the mark pattern 70, the gradation of the pixel 71 between the elements 77 becomes 255 gradation, and changes to the mark pattern 78 having the vertically long elements 77 as described above.

  Next, when the mark pattern 78 after this horizontal edge enhancement filter processing is further processed by the vertical edge enhancement filter, the gradation of the pixel 71 between the elements 77 in FIG. 8B in the horizontal direction of the mark pattern 78. Changes from 0 gradation to 255 gradation, and changes to a mark pattern 80 having a rectangular element 79 as shown in FIG.

  As described above, when the horizontal edge enhancement filter process is executed by reversing the order of steps S4 and S5 shown in FIG. 4, and then the vertical edge enhancement filter process is executed on the image after the horizontal edge enhancement filter process. In FIG. 8, the mark turn 80 of the image shown in FIG. 8C is searched by pattern matching executed in step S3 of FIG.

  FIG. 5 is a second flowchart specifically showing the contents of the combination of the vertical edge enhancement filter process and the horizontal edge enhancement filter process executed in steps S1 and S2 of FIG.

  In FIG. 5, after acquiring an image for inputting a template pattern image or a search target image, vertical edge enhancement filter processing is executed on the input image in step S <b> 6, and the template pattern image or search target is also determined in step S <b> 7. The horizontal edge enhancement filter process is executed on the input image. Next, in step S8, the pixel value of the pixel at the same place is compared with the pixel value of the output image for the image of the vertical edge enhancement filter processing result and the image of the horizontal edge enhancement filter processing result. Execute the process.

  FIG. 9 and FIG. 10 show the result of executing the filtering process by the method shown in FIG. 5 on the mark pattern in the search target image.

  FIG. 9 shows a mark pattern 74 having a horizontally long element 73 obtained by processing the mark pattern 70 shown in FIG. 7A using a vertical edge enhancement filter, and a mark pattern having a vertically long element 77 processed by a horizontal edge enhancement filter. 78 shows an image 82 superimposed on 78.

  In order to extract pixels that are common (overlapped) to both the mark pattern 74 and the mark pattern 78 from the image 82, a pixel value having a smaller pixel value is selected from the common pixels, and this is selected as the output image. When the pixel value processing (see step S8 in FIG. 5) is executed, a mark pattern 83 shown in FIG. 10 is obtained.

  When the filtering process is executed by the method shown in FIG. 5, the mark turn 83 of the image shown in FIG. 10 is searched by the pattern matching executed in step S3 of FIG.

  FIG. 11 shows a case where the search target image is processed by the vertical edge enhancement filter. In addition to the mark turn 85, the search target image 100 shown in FIG. 10A includes a pseudo pattern 86 that causes pseudo matching. In FIG. 11A, the mark pattern 85 is displayed as a rectangular pattern for convenience, but actually has the same shape as the mark pattern 70 shown in FIG. Note that the mark pattern 85, which is an image in the direction in which vertical edge enhancement filter processing is performed, is set smaller than the vertical edge enhancement filter processing area.

  When the search target image 100 is processed by the vertical edge enhancement filter, a search target image 101 shown in FIG. 11B is obtained. In the search target image 101, the pseudo pattern 86 changes to a vertical edge pattern 89, and the mark pattern 85 is a mark pattern 87 having the same shape as the mark pattern 74 shown in FIG. 7B (displayed as a rectangular pattern in the figure for convenience). To change.

  FIG. 12 illustrates a case where the horizontal edge enhancement filter process is further performed on the search target image 101 that has been subjected to the vertical edge enhancement filter process. Note that the mark pattern 87, which is an image in the direction in which the horizontal edge enhancement filter process is performed, is set smaller than the horizontal edge enhancement filter processing area.

  When the search target image 101 is processed by the horizontal edge enhancement filter, a search target image 102 shown in FIG. 12 is obtained. In this search target image 102, the vertical edge pattern 89 becomes an end point 90 whose end is changed to a point, and the mark pattern 87 is a mark pattern 88 having the same shape as the mark pattern 76 shown in FIG. (Displayed as a rectangular pattern for convenience).

  As described above, when the search target image 100 is processed by the vertical / horizontal edge emphasis filter, the pseudo pattern 86 that causes the pseudo matching finally becomes the end point 90, and the size and shape of the mark pattern 88 are greatly different. As a result, pseudo matching can be reduced and matching accuracy can be increased.

  Further, since the pseudo pattern 86 finally becomes the end point 90, the dimension of the mark pattern 88 can be set small.

  The present invention is not limited to the one shown in the above embodiment. Edge enhancement filter processing has been shown as image processing that extracts a portion that changes between adjacent pixels from the pixels that make up the template image and search target image, and removes other portions, but this is not the only case Is not to be done.

  Moreover, although the case where the process of both the vertical edge emphasis filter process and the horizontal edge emphasis filter process was shown was shown, you may perform only any one process. In short, as long as the pseudo patterns can be reduced, only one of the vertical edge enhancement filter processing and the horizontal edge enhancement filter processing may be performed.

It is a schematic block diagram which shows one Embodiment of the position detection apparatus of this invention. It is a block diagram which shows the content of the template image process part 10 (search target image process part 20) of the position detection apparatus of FIG. It is a flowchart which shows one Embodiment of the position detection method of this invention. FIG. 4 is a first flowchart specifically showing the content of a combination of vertical edge enhancement filter processing and horizontal edge enhancement filter processing executed in step S1 and step S2 of FIG. 3. FIG. 4 is a second flowchart specifically showing the content of a combination of vertical edge enhancement filter processing and horizontal edge enhancement filter processing executed in step S1 and step S2 of FIG. 3. It is explanatory drawing of the vertical edge emphasis filter and horizontal edge emphasis filter used for the vertical / horizontal edge emphasis filter process in the template image process part 10 and the search object image process part 20 of FIG. FIG. 5 is an explanatory diagram illustrating a result of performing filter processing on a mark pattern in a search target image by the method illustrated in FIG. 4. FIG. 5 is an explanatory diagram illustrating a result of performing filter processing on a mark pattern in a search target image by the method illustrated in FIG. 4. It is explanatory drawing which shows the result of having performed the filter process with the method shown in FIG. 5 with respect to the mark pattern in a search object image. It is explanatory drawing which shows the result of having performed the filter process with the method shown in FIG. 5 with respect to the mark pattern in a search object image. It is explanatory drawing at the time of processing with a vertical edge emphasis filter with respect to a search object image. It is explanatory drawing at the time of carrying out the horizontal edge emphasis filter process further with respect to the search object image which performed the vertical edge emphasis filter process.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Template process part 20 Search object image process part 30 Store part 40 Template matching process part 50 Control part 60 Kernel of vertical edge emphasis filter 61 Kernel of horizontal edge emphasis filter 70 Mark pattern 71 Pixel 72 Element (element pattern)
73 elements (element patterns)
77 elements (element patterns)
74 Mark pattern 76 Mark pattern 78 Mark pattern 80 Mark pattern 83 Mark pattern 85 Mark pattern 86 Pseudo pattern 87 Mark pattern 88 Mark pattern 100 Search target image 101 Search target image 102 Search target image

Claims (8)

An apparatus for detecting the position of a mark pattern that matches a template in a search target image,
A search target image processing unit that performs image processing to extract a portion that changes between adjacent pixels among the pixels constituting the image to the search target image;
A detection unit that detects a position of the mark pattern that matches a template image of the template in the search target image subjected to the image processing;
A position detection device comprising: The position detection device according to claim 1,
A template image processing unit that performs image processing for extracting a portion that changes between adjacent pixels among the pixels constituting the image to the template image,
The detection unit includes a position of the mark pattern that matches the template image subjected to the image processing by the template image processing unit in the search target image subjected to the image processing by the search target image processing unit. A position detecting device for detecting In the position detection device according to claim 1 or 2,
The position detection apparatus according to claim 1, wherein the image processing is processing for performing edge enhancement filter processing in two directions intersecting with each other. In the position detection device according to claim 3,
The position detection apparatus according to claim 1, wherein the mark pattern includes a plurality of element patterns, and the element patterns are smaller than a processing area of the edge enhancement filter processing. The position detection device according to claim 4,
The position detection apparatus according to claim 1, wherein the element pattern is an image having a corresponding number of pixels of 2 × 2. In the position detection device according to claim 3,
In the edge enhancement filter processing, the edge enhancement filter processing is performed on the image in one direction out of the two directions, and then the edge enhancement filter processing is performed on the image subjected to the edge enhancement filter processing in this one direction in the other direction. A position detection device characterized by performing a filtering process. In the position detection device according to claim 3,
In the edge filter processing, the edge enhancement filter processing is performed on the image in one direction out of the two directions, the edge enhancement filter processing is performed on the image in the other direction, and the edge enhancement filter processing is performed on the image. A position detecting device, wherein both images are overlapped and the overlapped portion is used as the image. A method of detecting a position of a mark pattern that matches a template in a search target image,
A template processing step of applying edge enhancement filter processing to the template image of the template in two directions intersecting each other;
A search target image processing step for subjecting the search target image to edge enhancement filter processing in two directions intersecting each other;
Detecting the position of the mark pattern that matches the template image subjected to the image processing in the search target image subjected to the image processing;
A position detection method comprising:

Images