JP3992867B2 - Position detecting method and apparatus, and recording medium - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a position detection method and apparatus for detecting the position and / or size of a substantially line-symmetric object included in an observation image, and a computer-readable record in which a program for causing a computer to execute the position detection method is recorded. It relates to the medium.
[0002]
[Prior art]
In a detection apparatus for inspecting a fine pattern on a wafer or a reticle in manufacturing or inspection of semiconductor devices such as VLSI, a square box of about 10 μm is used as an inspection object on a square box of about 20 μm manufactured in the previous process. What is manufactured and inspects the alignment accuracy by detecting the edge position of each box is known. In such an inspection apparatus, paying attention to the fact that the inspection target is substantially line symmetric, a first predetermined region is set on one side of the line symmetry of the observation image representing the inspection target, and the set first An image of one predetermined area is inverted to set a reference image related to the first predetermined area, a second predetermined area is set on the other side of line symmetry with respect to the set reference image, and the first A method for detecting the position of the inspection object from the center position of the position of the second predetermined region has been proposed (Japanese Patent Laid-Open No. 6-258045). According to this method, when detecting the position of the target pattern having symmetry from the observation image data of the inspection target, the other edge is based on the reference image of the first predetermined area representing the one edge pattern. The pattern can be determined, and the center position of both edge patterns, that is, the alignment of the inspection object can be accurately detected regardless of the image state of the inspection object.
[0003]
[Problems to be solved by the invention]
The method described in Japanese Patent Laid-Open No. 6-258045 can detect the position of a substantially line-symmetric object with high accuracy. For example, although the object is substantially line-symmetric like a human face image, If the symmetry is lost in the details, the position of the object cannot be detected accurately.
[0004]
The present invention has been made in view of the above circumstances, and is capable of accurately detecting the position and / or size of an object even when the symmetry is broken in the details of the object, although it is substantially line symmetric. An object of the present invention is to provide a position detection method and apparatus that can be used, and a computer-readable recording medium that records a program for causing a computer to execute the position detection method.
[0005]
[Means for Solving the Problems]
The position detection method according to the present invention is a position detection method for detecting the position and / or size of an object when the object included in the observation image is substantially line symmetric.
A first predetermined region is set on one side of the object in line symmetry;
Reversing an image other than the specific area in the first predetermined area to set a reference image related to the first predetermined area;
Based on the reference image, a second predetermined region is set on the other side of the line symmetry,
The position and / or size of the object is detected based on the positions of the first and second predetermined regions.
[0006]
Here, the “reference image” is created by inverting an area other than the specific area in the first predetermined area.
[0007]
In addition, the “specific area” refers to an area in which symmetry in the object is impaired in the first predetermined area.
[0008]
Further, “invert” means to invert an image other than the specific region so as to be line symmetric with respect to an axis parallel to the substantially line symmetric axis of the object included in the observation image.
[0009]
In addition, in order to prevent the image from being lost due to the inversion of the image outside the specific area, the shape of the specific area is set to be substantially line symmetric with respect to the symmetry axis when the image other than the specific area is inverted. It is preferable to do.
[0010]
An image processing device according to the present invention is a position detection device that detects the position and / or size of an object when the object included in the observation image is substantially line symmetric.
First setting means for setting a first predetermined region on one side of line symmetry of the object;
A reference image setting means for inverting an image other than the specific area in the first predetermined area and setting a reference image related to the first predetermined area;
Second setting means for setting a second predetermined region on the other side of the line symmetry based on the reference image;
And detecting means for detecting the position and / or size of the object based on the positions of the first and second predetermined regions.
[0011]
In addition, you may record and provide on a computer-readable recording medium as a program for making a computer perform the position detection method by this invention.
[0012]
【The invention's effect】
According to the present invention, when detecting the position and / or size of a substantially line-symmetric object included in the observation image, the inside of the first predetermined region set on one side of the object's line symmetry A reference image is set by inverting an image other than the specific area in FIG. 2, and a second predetermined area is set on the other side of the line symmetry based on the reference image. Here, even if there is a portion in which the symmetry is broken with respect to the other side of line symmetry in the first predetermined region, if the portion in which the symmetry is broken is set as a specific region, the symmetry is reduced. The second predetermined area can be set at a position substantially symmetrical with the first predetermined area without being affected by the collapsed portion. Therefore, the position and / or size of the object can be accurately detected from the positions of the first and second predetermined regions.
[0013]
DETAILED DESCRIPTION OF THE INVENTION
Embodiments of the present invention will be described below with reference to the drawings.
[0014]
FIG. 1 is a schematic block diagram showing a configuration of a position detection device according to an embodiment of the present invention. As shown in FIG. 1, the position detection device 1 according to the present embodiment detects the position and / or size of a substantially line-symmetric object included in an image represented by image data S acquired by the imaging means 2. Based on the image data S, a first predetermined area is set on one side of the line symmetry of the object, and an image other than the specific area in the first predetermined area is inverted. Based on the first setting means 3 for setting the obtained reference image relating to the first predetermined area as the template T and the template T set in the first setting means 3, the second image is set on the other side of the line symmetry. Second setting means 4 for setting the predetermined area, and detection means 5 for detecting the position and / or size of the object based on the first and second predetermined areas.
[0015]
In the present embodiment, the object is a human face image, and the position and size of the face image are detected based on the image data S acquired by imaging the human face. Things are not limited to face images.
[0016]
In the first setting means 3, the template T is set as follows. FIG. 2A is a diagram showing a human face image represented by the image data S. FIG. The face image shown in FIG. 2A has symmetry with respect to an axis (broken line) parallel to the Y axis, and the eyes are directed to the left side (right side when viewed from a person). In the first setting means 3, first, a filtering process is performed on one side of line symmetry (the left side as viewed in FIG. 2) by a filter for detecting a light and dark pattern as shown in FIG. As shown in (b), a first predetermined area A1 including the left eye is set. In the present embodiment, since the face image whose position and size are detected does not change significantly, for example, based on the average size of the face image, the filter shown in FIG. Is set. Note that the use of the filter shown in FIG. 3B can cope with the case where the eyes are facing the right side.
[0017]
Further, as shown in FIG. 4, a specific area N1 including a black eye is set at the center in the first predetermined area A1, and the area N2 other than the specific area N1 is inverted with an axis parallel to the Y axis as an axis of symmetry. Then, by combining with the specific area N1, a template T is created as shown in FIG.
[0018]
The second setting means 4 first sets a search area E sufficiently larger than the template T on the other side of the line symmetry (right side as viewed in FIG. 2), and scans the template T in the search area E. Thus, the cross-correlation coefficient is obtained, the area having the largest correlation coefficient is detected, and this area is set as the second predetermined area A2 as shown in FIG.
[0019]
The detecting means 5 calculates center coordinates X1 and X2 in the X-axis direction of the first predetermined area A1 and the second predetermined area A2, and determines the position and size of the face image based on the center coordinates X1 and X2. It is what you want. Specifically, the midpoint of the center coordinates X1 and X2 is regarded as the midpoint of both eyes, and this is regarded as the position of the face image, and the difference value is regarded as the distance between both eyes, which is obtained as the size of the face image (FIG. 2 ( d)).
[0020]
Next, the operation of this embodiment will be described. FIG. 6 is a flowchart showing the operation of this embodiment.
[0021]
First, the image data S is acquired by the imaging means 2 (step S1). Next, a filtering process is performed on the left area of FIG. 2A by the filter shown in FIG. 3A to set a first predetermined area A1 (step S2). Next, the center coordinate in the X-axis direction of the set first predetermined area A1 is calculated as X1 (step S3).
[0022]
Then, as shown in FIG. 4, an area N2 representing the specific area N1 and a part other than the specific area N1 is set in the first predetermined area A1 (step S4). Further, as shown in FIG. 5, a template T is created by inverting the region N2 with the axis parallel to the Y axis as the axis of symmetry (step S5).
[0023]
Next, in the area on the right side of FIG. 2A, a search area E that is sufficiently larger in the X-axis direction than the template T is set (step S6). Then, in the search area E, template matching is performed to obtain a correlation coefficient between the template T and a region corresponding to the template T in the search area E while shifting the template T pixel by pixel in the X-axis direction and the Y-axis direction ( Step S7).
[0024]
Next, the region of the template T where the calculated correlation coefficient is the maximum value is set as the second predetermined region A2 (step S8), and the center position of the second predetermined region A2 in the X-axis direction is determined as X2 ( Step S9). Then, the center coordinate Xc between the coordinate X1 obtained in step S3 and the coordinate X2 obtained in step S9 is obtained as the position of the face image by the calculation of Xc = (X1 + X2) / 2, and the coordinate X2 and the coordinate X1 The difference value ΔX is obtained as the size of the face image (step S10), and the process ends.
[0025]
As described above, according to the present embodiment, even in the case where only the eyes shown in FIG. 2 are face-to-face images, the lines are symmetrical but the symmetry is broken in the details. Since the second predetermined area A2 can be set at a position symmetrical to the first predetermined area A1 without being affected by the portion where the characteristics are broken, the first and second predetermined areas A1, A2 The position and size of an object such as a face image can be accurately detected from the center position.
[0026]
In addition, since the position and size of the face image can be obtained with high accuracy, it is possible to cut out the person with a stable position and size from a large number of images in which the person is photographed.
[0027]
In the above embodiment, both the position and the size of the face image are obtained, but only one of them may be obtained.
[0028]
In the above embodiment, the specific region N1 is set at the center of the first predetermined region A1, and the template T is created by inverting the region N2 other than the specific region N1, but the target image In accordance with the state in which the symmetry of the predetermined area A1 is impaired, the area other than the central part of the first predetermined area A1 is set as the specific area N1, and the template N is inverted by inverting the area N2 other than the specific area N1. You may create it.
[0029]
Furthermore, in the above-described embodiment, after setting the first predetermined area A1, the template T is created and template matching is performed two-dimensionally to set the second predetermined area A2. If only the left and right positions need to be obtained, the processing may be performed only on the left and right one-dimensional data as follows.
[0030]
FIG. 7 is a flowchart showing another operation of the present embodiment. First, the image data S is acquired in step S11, and the first predetermined area A1 is set in step S12. Next, in step S13, line data D [i] (i: 0 to (A-1) in the X axis direction within the range of the width Y3 in the Y axis direction in the first predetermined area A1 as shown in FIG. A is the number of data).
[0031]
Then, in the line data D [i], the data in the first predetermined area A1 is cut out as a template T (step S14). At this time, if the X coordinate of the start point of the first predetermined area A1 is X3 and the number of data of the template T is N, the cut out data T1 [j] (j: 1 to (N−1)) is
T1 [j] = D [X3 + (N-1) -j]
It becomes. The data T1 [j] is obtained by inverting the first predetermined area A1 with an axis parallel to the Y axis as an axis of symmetry.
[0032]
Next, it is determined whether or not the line data D [i] is within the range of the width Y4 in the Y-axis direction in the region N1 shown in FIG. 8 (step S15). If not, the process proceeds to step S17. . If it is within the range, the data corresponding to the region N1 is inverted (step S16). Here, if the X coordinate of the start point of the region N1 is X4 and the number of data is q, the data is T2 [k] (k: 0 to q-1),
T2 [k] = D [X4 + k]
It becomes.
[0033]
In step S17, a search area E that is sufficiently larger than the template T is set. If the X coordinate of the starting point is X5 and the number of data is L (N <L), the data E [k] (k: 0 to L-1) in the search area E is
E [k] = D [X5 + k]
It becomes. Next, the initial value of m is set to 0 (step S18), and data is cut out from the search area E by the number N of data in step S19. Data Em [j] (j: 1 to (N-1)) cut out at this time is
Em [j] = E [m + j]
It becomes.
[0034]
In step S20, a cross-correlation coefficient between Em [j] and T1 [j] is obtained and set as C [m]. In steps S21 and S22, the processes of steps S19 and S20 are performed while changing m from 0 to (L−N).
[0035]
Then, the value of m that is the maximum value of the correlation coefficient C [m] is obtained, and this is set as Xmax (step S23). In step S24, the center coordinates Xcnt of both eyes are calculated by the following formula.
[0036]
Xcnt = (X3 + N-1 + X5 + Xmax) / 2
Thereby, the X coordinate of the center of both eyes in one line data D [i] can be obtained.
[0037]
Then, the above processing is performed on all the line data within the range of Y3 shown in FIG. 8 to obtain the center coordinate Xcnt for each line, and the average value thereof is obtained as the position of the face image.
[0038]
As described above, the position of the face image can also be obtained by the processing shown in FIG.
[Brief description of the drawings]
FIG. 1 is a schematic block diagram showing a configuration of a position detection apparatus according to an embodiment of the present invention. FIG. 2 is a diagram showing a face image. FIG. 3 is a diagram showing a filter for setting a first predetermined region. FIG. 5 is a diagram showing a template. FIG. 6 is a flowchart showing the operation of the present embodiment. FIG. 7 is a flowchart showing another operation of the embodiment. FIG. Illustration for explanation [Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 Position detection apparatus 2 Imaging means 3 1st setting means 4 2nd setting means 5 Detection means

Claims (3)

A position detection method for detecting a position and / or size of an object when the object included in the observation image is substantially line symmetric,
A first predetermined region is set on one side of the object in line symmetry;
Reversing an image other than the specific area in the first predetermined area to set a reference image related to the first predetermined area;
Setting a second predetermined region on the other side of the line symmetry based on the reference image;
A position detection method comprising detecting the position and / or size of the object based on the positions of the first and second predetermined regions. A position detection device for detecting the position and / or size of an object when the object included in the observation image is substantially line symmetric,
First setting means for setting a first predetermined region on one side of line symmetry of the object;
A reference image setting means for inverting an image other than the specific area in the first predetermined area and setting a reference image related to the first predetermined area;
Second setting means for setting a second predetermined region on the other side of the line symmetry based on the reference image;
A position detection apparatus comprising: a detection unit configured to detect a position and / or size of the object based on positions of the first and second predetermined regions. A computer-readable recording medium storing a program for causing a computer to execute a position detection method for detecting the position and / or size of an object when the object included in the observation image is substantially line symmetric There,
A procedure for setting a first predetermined region on one side of line symmetry of the object;
A procedure for inverting an image other than the specific area in the first predetermined area and setting a reference image related to the first predetermined area;
Setting a second predetermined region on the other side of the line symmetry based on the reference image;
A computer-readable recording medium comprising: a procedure for detecting the position and / or size of the object based on the positions of the first and second predetermined areas.

Images