JPH09113225A - Image processing device and method and device for visual inspection of object - Google Patents

Abstract

PROBLEM TO BE SOLVED: To perform the visual inspection of a circular object in high accuracy by using a picture wherein the circular object having a concentric repeating pattern is arranged at unfixed position and rotational angle. SOLUTION: In a model preparation section 4, a correction model for positional displacement as a picture of an area including the peripheral part of a reference object, a correction model for rotational displacement as a picture of an area including an optional unit pattern, and an inspection model as a picture of inspection area are prepared in advance by using a reference picture, and an area corresponding to the correction model for positional displacement is extracted from a picture to be inspected through a part 7 for correcting positional displacement, then the positional displacement quantity of an object to be inspected is obtained according to the position of the extracted area, for correction of positional displacement. In addition, an area corresponding to the correction model for rotational displacement is extracted by a part 8 for correcting rotational displacement, and the rotational displacement quantity is obtained according to the position of the extracted area, for correction of positional displacement, then an inspection execution part 9 performs the visual inspection by comparing the picture of the inspection area in the corrected picture to be inspected with the inspection model.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a plurality of characteristic unit patterns arranged concentrically (hereinafter, the plurality of unit patterns are simply referred to as concentric circular repeating patterns).
An image processing device and method for arranging the circular object whose position and rotation angle on the image are unspecified at a predetermined position and a predetermined rotation position on the image,
In addition, the present invention relates to an object appearance inspection device for inspecting the presence or absence of a defect on the appearance of the circular object using the image.

[0002]

2. Description of the Related Art An object appearance inspection apparatus for inspecting a circular product (object to be inspected) having a concentric repeating pattern for appearance defects is an image (model) of a product (reference object) having no defect as a reference. ) Is registered in advance, the image of the object to be inspected (inspection image) is compared with the model (pattern matching process), and the presence or absence of a defect is determined by the degree of coincidence.

By the way, a camera for picking up an image of an object to be inspected is usually arranged on the product unloading side of a manufacturing line, and the object to be inspected is randomly placed on a carrier and moves, so that the object to be inspected is picked up. In the image to be inspected obtained in this way, the position and rotation angle of the object to be inspected are unspecified.

Therefore, the following is an example of a conventional object appearance inspection apparatus.

An image of the reference object is cut out from a reference image obtained by picking up an image of the reference object (the angle of view of which is the same as the image to be inspected) to obtain a reference model. Create each rotated image by using as a rotation model, register these models, and perform pattern matching processing with each area in the image to be inspected (the angle of view is the same as the model) in sequence. If there is a combination of the region of the image to be inspected and the model whose degree of coincidence is a certain value or more, it is determined that there is no defect in the appearance of the object to be inspected.

The above rotation model is created within a range of ± (360 / (2N)) ° when the number of unit patterns is N.

However, in the above-mentioned conventional object appearance inspection apparatus, since a plurality of models (reference model and rotation model) consisting of a large image of the entire circular object must be stored, a huge memory capacity is required. Moreover, since the pattern matching process with an arbitrary region of the image to be inspected must be performed using this large model, this process takes too much time, which is not practical.

Therefore, the following is an example of a conventional object appearance inspection apparatus which copes with such a problem.

In the reference image, a first area including a unit pattern of the reference object and a second area including another unit pattern are set, and the images of the first and second areas are cut out to obtain the first area and the first area, respectively. And a second reference model for correction, and a rotation model for correction is created in a range of ± (360 / (2N)) ° with respect to the first and second reference models for correction. By registering the second correction model and performing pattern matching processing between an arbitrary region of the inspection target image and each of the first and second correction models, in the inspection target image, the first and the second correction models are registered. Extract the corresponding areas of the two areas, respectively, based on the positional relationship between the two extracted areas, calculate the position and rotation angle of the object to be inspected, the amount of displacement from the reference position and the reference rotation angle The amount of rotation deviation is obtained, and image processing for correcting these deviations is performed on the inspection target image.

In the inspection object image whose displacement amount has been corrected, the concentric repeating pattern is arranged at a predetermined position, and therefore the inspection object is arranged at the reference position and the reference rotation position, so that the predetermined area including the inspection object is located. Is set as an inspection area in advance, and an image of an area including a reference object is cut out from the reference image and registered as an inspection model (a rotation model is not necessary). A pattern matching process with the image of the inspection area in the target image is performed.

[0011]

However, in such a conventional object appearance inspection apparatus, when the number of unit patterns in the object to be inspected is large, the correction accuracy of the deviation amount is deteriorated, so that the defect of the object to be inspected There was a problem that the detection accuracy deteriorates.

For example, if the number of unit patterns of a circular object is 2
In the reference image, the area including a certain unit pattern is set as the first area in the reference image, and the area including the tenth unit pattern counted from the unit pattern of the first area is set as the second area. If set, in the image to be inspected, the area including the ninth or eleventh unit pattern counted from the extracted unit pattern of the area corresponding to the first area is the area corresponding to the second area. May be erroneously extracted, and the calculated deviation amount includes an error, so that the concentric repeating pattern of the inspection object is not corrected to a predetermined position.

The present invention solves such a conventional problem, and in an image of a circular object having a concentric circular repeating pattern, the circular object whose position and rotation angle are not specified is rotated at a predetermined position and a predetermined rotation. By using an image processing device and method capable of arranging the circular object with high accuracy, and an image in which the circular object is arranged at an unspecified position and rotation angle, it is possible to determine whether or not there is a defect in the appearance of the circular object. It is an object of the present invention to provide an object appearance inspection device that can inspect accurately.

[0014]

In order to achieve the above object, the image processing method according to claim 1 of the present invention is a reference object for a circular object in which a plurality of characteristic unit patterns are concentrically arranged. The reference position and the reference rotation position of the object are calculated based on the reference image obtained by capturing and are stored in advance, and the processing target image is obtained by capturing the processing target object. A method of arranging at the reference position and the reference rotation position, wherein the eigenpoint of the reference object is set based on the position of at least one misregistration correction region including the peripheral portion of the reference object, which is set on the reference image. The coordinates are calculated, and the eigenpoint coordinates are stored as the reference position, and the image of the misregistration correction area is cut out from the reference image and stored in advance as a misregistration correction model. Place was set on the reference image,
The rotation angle of the reference object is calculated based on the positional relationship between the rotation deviation correction area including the arbitrary unit pattern and the reference position, and the rotation angle is stored as the reference rotation position, and the rotation deviation correction is performed. The image of the working area is cut out from the reference image and stored in advance as a rotation deviation correction model, and in the processing target image, an area corresponding to the position deviation correction model is extracted, and based on the position of the extracted area. Calculate the eigenpoint coordinates of the object to be processed, determine the amount of displacement of the object to be processed from the difference between the coordinates of the eigenpoint of the object to be processed and the reference position, and scroll predetermined to the image to be processed By performing the processing, the positional deviation amount is corrected, and the region corresponding to the rotational deviation correction model is extracted from the processing target image subjected to the positional deviation correction. Calculating a rotation angle of the object to be treated on the basis of the extracted region with the positional relationship between the reference position,
From the difference between the rotation angle of the object to be processed and the reference rotational position, the amount of rotation deviation of the object to be processed is obtained, and the predetermined image is subjected to a predetermined affine transformation process to obtain the rotation. This is a method characterized by correcting the amount of deviation.

An image processing method according to a second aspect of the present invention is the image processing method according to the first aspect, in which two unit patterns that include the rotation deviation correction area and are adjacent to this area are adjacent to each other. A search area set on the reference image, which includes each of the side halves, is stored in advance, and an area corresponding to the rotation deviation correction model is extracted from the search area in the processing target image having the positional deviation amount corrected. It is a method characterized by that.

Next, the image processing apparatus according to claim 3 of the present invention is based on a reference image obtained by imaging a reference object for a circular object in which a plurality of characteristic unit patterns are concentrically arranged. A reference position and a reference rotation position of an object are calculated and stored in advance, and in the processing target image obtained by imaging the processing target object, the processing target object is arranged at the reference position and the reference rotation position. And an operation unit for setting at least one position deviation correction area including the peripheral portion of the reference object on the reference image, and setting a rotation deviation correction area including any unit pattern. , Calculating the eigenpoint coordinates of the reference object as the reference position based on the position of the misregistration correction region, and based on the positional relationship between the rotation misregistration correction region and the reference position A reference value calculating unit that calculates the rotation angle of the reference object as the reference rotation position, and a position deviation correction model is created by cutting out the image of the position deviation correction region from the reference image, and the rotation deviation is also generated. A model creating unit that creates a rotation deviation correction model by cutting out an image of the correction area, a model storage unit that stores the position deviation correction model and the rotation deviation correction model, and the reference position and the reference rotation position are stored. A reference value storage unit to be stored, and in the processing target image, an area corresponding to the misregistration correction model is extracted,
Position deviation amount calculating means for calculating the eigenpoint coordinates of the processing target object based on the position of the extracted region, and for calculating the position deviation amount of the processing target object from the difference between the eigenpoint coordinates and the reference position; By performing a predetermined scrolling process on the processing target image, a positional shift correction unit that corrects the positional shift amount and an area corresponding to the rotational shift correction model in the positional shift corrected processing target image are generated. The rotation angle of the object to be processed is calculated based on the positional relationship between the extracted area and the reference position, and the amount of rotational deviation of the object to be processed is calculated from the difference between the rotation angle and the reference rotational position. A rotation deviation amount calculating means and a rotation deviation for correcting the rotation deviation amount of the object to be processed by performing a predetermined affine transformation process on the position deviation-corrected processing object image. It is characterized in that it comprises a positive means.

An image processing apparatus according to a fourth aspect of the present invention is the image processing apparatus according to the third aspect, wherein the image processing apparatus includes two areas that include the rotation deviation correction area and that are adjacent to the rotation deviation correction area. The operation means for setting a search area including the adjacent halves of the unit pattern on the reference image, and area storage means for storing the search area,
The rotation deviation amount calculation means for extracting an area corresponding to the rotation deviation correction model in the search area in the position deviation corrected image to be processed is provided.

Next, in the object appearance inspection apparatus according to the fifth aspect of the present invention, with respect to a circular object in which a plurality of characteristic unit patterns are concentrically arranged, an image of the object to be inspected is stored in advance as a reference. By comparing with an image of the object, the presence or absence of a defect on the appearance of the inspection target object is inspected, and a peripheral portion of the reference object is set on a reference image obtained by imaging the reference object. To set at least one position deviation correction area including the unit, and to set a rotation deviation correction region including any unit pattern, and to set at least one inspection area including all or part of the reference object. Of the operation means and the position of the misregistration correction area to calculate the eigenpoint coordinates of the reference object as a reference position, and the positional relationship between the rotation misregistration correction area and the reference position is calculated. Then, a reference value calculating means for calculating the rotation angle of the reference object as a reference rotational position, and a position deviation correction model is created by cutting out the image of the position deviation correction region from the reference image, and the rotation is also changed. A model creating unit that creates a rotation misalignment correction model by cutting out an image of the misregistration correction region, and further creates an inspection model by cutting out the image of the inspection region, and an area storage unit that stores the inspection region. A model storage unit that stores the positional deviation correction model, the rotational deviation correction model, and the inspection model; a reference value storage unit that stores the reference position and the reference rotational position; and an image of the inspection target object. In the obtained inspection target image,
An area corresponding to the misalignment correction model is extracted, eigenpoint coordinates of the inspection object are calculated based on the position of the extracted area, and the inspection object is calculated from the difference between the eigenpoint coordinates and the reference position. Position deviation amount calculating means for calculating the position deviation amount, position deviation correcting means for correcting the position deviation amount by performing a predetermined scrolling process on the inspection object image, and the position deviation corrected inspection object In the image, a region corresponding to the rotation deviation correction model is extracted, the rotation angle of the inspection target object is calculated based on the positional relationship between the extracted region and the reference position, and the rotation angle and the reference rotation position are calculated. By performing a predetermined affine transformation process on the inspection target image whose positional deviation has been corrected, By comparing a rotation deviation correction unit that corrects the rotation deviation amount, an image of the inspection region in the inspection target image in which the positional deviation and the rotation deviation have been corrected, and the inspection model, the inspection target object An inspection execution means for inspecting the presence or absence of a defect in appearance is provided.

An object appearance inspection apparatus according to a sixth aspect of the present invention is the object appearance inspection apparatus according to the fifth aspect, which includes the rotation deviation correction area and is adjacent to the rotation deviation correction area. The operation means for setting a search area including two adjacent side halves of two unit patterns on the reference image, the area storage means for storing the search area, and the position-corrected inspection target image The rotation deviation amount calculating means for extracting an area corresponding to the rotation deviation correction model in the search area is provided.

Therefore, according to the image processing method of the first aspect and the image processing apparatus of the third aspect of the present invention, at least one positional deviation including the peripheral portion of the reference object set on the reference image by the operating means. Based on the position of the correction area, the eigenpoint coordinates of the reference object are calculated by the reference value calculating means, the eigenpoint coordinates are stored as the reference position in the reference value storing means, and the positional deviation is corrected by the model creating means. A position misalignment correction model is created by cutting out the image of the working area from the reference image and stored in advance in the model storage means, and rotation misregistration correction including an arbitrary unit pattern set on the reference image by the operation means is performed. The rotation angle of the reference object is calculated by the reference value calculation means based on the positional relationship between the working area and the reference position, and this rotation angle is used as the reference rotation position. Is stored in the reference value storage means, and a model for the rotation deviation correction is cut out from the reference image by the model creation means to create a rotation deviation correction model and stored in advance in the model storage means. A region corresponding to the displacement correction model is extracted from the image to be processed by the amount calculation means, the eigenpoint coordinates of the object to be processed are calculated based on the position of the extracted region, and the eigenpoint of the object to be processed is calculated. From the difference between the coordinates and the reference position, obtain the amount of positional deviation of the object to be processed,
By performing a predetermined scrolling process on the processing target image by the positional deviation correction means, the positional deviation amount is corrected, and by the rotational deviation amount calculation means, the positional deviation correction is performed on the processing target image. A region corresponding to the model is extracted, the rotation angle of the processing target object is calculated based on the positional relationship between the extracted region and the reference position, and processing is performed from the difference between the rotation angle of the processing target object and the reference rotation position. By obtaining the amount of rotation deviation of the target object and performing a predetermined affine transformation process on the processing target image whose position deviation has been corrected by the rotation deviation correction means, the amount of rotation deviation is corrected to form a concentric circular repeating pattern. Arrangement of the circular object, whose position and rotation angle are unspecified, at a predetermined position and a predetermined rotation position with high accuracy in the image of the circular object Rukoto can.

According to the image processing method according to claim 2 and the image processing device according to claim 4 of the present invention, in the image processing method according to claim 1 and the image processing device according to claim 3 of the present invention, A search area, which is set on the reference image by the operation means, includes the rotation deviation correction area and includes two adjacent side halves of two unit patterns adjacent to this area, is stored in advance in the area storage means, and is rotated. A circular object having a concentric circular repeating pattern is determined in a short time by extracting a region corresponding to the rotational displacement correction model in the search region in the processing target image whose positional displacement amount has been corrected by the displacement amount calculating means. It is possible to highly accurately dispose the position and the predetermined rotation position.

Next, according to the object appearance inspection apparatus of the fifth aspect of the present invention, at least one misregistration correction area including the peripheral portion of the reference object is set on the reference image by the operating means, and A rotation deviation correction area including an arbitrary unit pattern is set, at least one inspection area including all or a part of the reference object is set, and the inspection area is stored in advance in the area storage means. The reference value calculating means calculates the eigenpoint coordinates of the reference object as a reference position on the basis of the position of the position deviation correction area, and further, based on the positional relationship between the rotation deviation correction area and the reference position, The rotation angle of the reference object is calculated as the reference rotation position, the calculated reference position and the reference rotation position are stored in the reference value storage means in advance, and the reference image is stored by the model creating means. Creates a positional deviation correction model by cutting the image of the positional deviation correction region,
Also, a rotation deviation correction model is created by cutting out the image of the rotation deviation correction area, and an inspection model is created by cutting out the image of the inspection area, and the created position deviation correction model and rotation deviation correction model are created. For model,
Further, the inspection model is stored in advance in the model storage means, and the position shift amount calculation means extracts a region corresponding to the position shift correction model in the inspection target image, and the inspection is performed based on the position of the extracted region. The eigenpoint coordinates of the target object are calculated, the displacement amount of the inspection target object is obtained from the difference between the eigenpoint coordinates and the reference position, and the predetermined displacement process is performed on the inspection target image by the displacement correction means. The positional deviation amount is corrected by the rotational deviation amount calculating means, and the region corresponding to the rotational deviation correction model is extracted from the inspection target image subjected to the positional deviation correction by the rotational deviation amount calculating means. The rotation angle of the object to be inspected is calculated based on the positional relationship between the object and the rotation deviation of the object to be inspected from the difference between this rotation angle and the reference rotation position The inspection target image in which the positional deviation is corrected by the corrector is subjected to a predetermined affine transformation process to correct the rotational deviation amount, and the inspection execution means corrects the positional deviation and the rotational deviation. By comparing the image of the inspection region in the image with the inspection model, the presence or absence of a defect on the appearance of the inspection target object is inspected, so that the circular object having the concentric circular repeating pattern has an unspecified position and By using the images arranged at the rotation angle, it is possible to inspect the appearance of the circular object with high accuracy.

According to a sixth aspect of the present invention, there is provided the object visual inspection apparatus according to the fifth aspect, which includes the rotation deviation correction area set on the reference image by the operating means. Further, a search area including two adjacent side halves of two unit patterns adjacent to this area is stored in advance in the area storage means, and the search area in the processing target image whose position deviation amount is corrected by the rotation deviation amount calculation means. In, by extracting a region corresponding to the rotation deviation correction model, by using an image in which the circular object is arranged at an unspecified position and a rotation angle, on the appearance of the circular object having a concentric repeating pattern The presence or absence of defects can be inspected in a short time with high accuracy.

[0024]

1 is a block diagram showing the configuration of an embodiment of an object appearance inspection apparatus of the present invention.

In FIG. 1, reference numeral 1 has a video camera or the like, and in a visual inspection of a circular object having a concentric repeating pattern, a reference object is obtained by imaging a reference object as a comparison reference, and an object to be inspected. Is an imaging device that obtains an inspection target image by capturing an image of the inspection target object.

An image memory 2 has an image memory and the like, and stores the reference image and the inspection target image, and also stores various images such as a model created by cutting out the reference image and the displacement-corrected inspection target image. It is a department.

Reference numeral 3 denotes a keyboard, a mouse, etc., on which the operator shifts a rotation displacement including any one of a displacement correction model region including the peripheral portion of the reference object and a unit pattern of the reference object. A correction model area, a search area including the rotational deviation correction model area, and a search area including two adjacent side halves of two unit patterns adjacent to the rotation deviation correction model area, and at least the whole or a part of the reference object It is an operation unit for setting one inspection area, specifying an operation mode, and inputting various commands.

Reference numeral 4 indicates a model for position deviation correction by cutting out an image of each position deviation correction model region from the reference image, and a model for position deviation correction by cutting out an image of the rotation position correction model region. Create
Further, the model creating unit creates an inspection model by cutting out the image of the inspection region.

Reference numeral 5 calculates the eigenpoint coordinates of the reference object as the reference position on the basis of the position of the displacement correction model area,
The reference value calculation unit calculates the rotation angle of the reference object as the reference rotation position based on the positional relationship between the rotation deviation correction model area and the reference position.

Reference numeral 6 is a data storage unit for storing various data such as a reference position, a reference rotational position, a search area, an inspection area, and an inspection result.

In the image to be inspected, 7 extracts an area that is pattern-matched with the model for positional deviation correction, calculates the eigenpoint coordinates of the object to be inspected based on the position of the extracted area, and the eigenpoint coordinates and the above-mentioned This is a position shift correction unit that corrects the position shift amount by obtaining the position shift amount of the inspection target object from the difference from the reference position and performing a predetermined scrolling process on the inspection target image.

Reference numeral 8 denotes an area to be pattern-matched with the rotational deviation correction model in the search area on the image to be inspected whose positional deviation has been corrected, and the object to be inspected based on the positional relationship between the extracted area and the reference position. By calculating the rotation angle of the object, obtaining the amount of rotation deviation of the inspection object from the difference between this rotation angle and the reference rotation position, and applying a predetermined affine transformation process to the inspection object image whose position deviation has been corrected. A rotation deviation correction unit that corrects the rotation deviation amount.

Reference numeral 9 is an inspection for inspecting the presence or absence of a defect on the appearance of the inspection object by comparing the image of the inspection area in the inspection image in which the positional deviation and the rotational deviation have been corrected with the inspection model. It is an execution unit.

Reference numeral 10 denotes a reference image, which has a display and the like.
The display unit displays various images such as an image to be inspected, a model, and the like, and reference position, reference rotational position, area setting data, displacement amount, rotation displacement amount, inspection result, and the like.

Reference numeral 11 denotes the image pickup device 1, the image storage unit 2, the model creation unit 4, the reference value calculation unit 5, the data storage unit 6, the positional deviation correction unit 7, the rotational deviation correction unit 8, the inspection execution unit 9, and the display unit. It is a control unit that integrally controls 10.

The image storage unit 2 corresponds to a model storage unit, the operation unit 3 corresponds to an operation unit, and the model creating unit 4 and the control unit 11 form a model creating unit.

The reference value calculation section 5 and the control section 11 constitute a reference value calculation means, and the data storage section 6 corresponds to the reference value storage means and the area storage means.

The positional deviation correcting section 7 and the control section 11 constitute a positional deviation amount calculating means and a positional deviation correcting means, and the rotational deviation correcting section 8 and the control section 11 constitute a rotational deviation amount calculating means and a rotational deviation correcting means. .

The inspection execution unit 9 and the control unit 11 constitute inspection execution means.

FIG. 2 is an external view showing an example of a circular object having a concentric circular repeating pattern, which is inspected by an object appearance inspection apparatus.

In FIG. 2, reference numeral 21 denotes the circular object, 2
Reference numeral 2 denotes a concentric circular repeating pattern formed on the circular object 21, and 22-1 and the like denote unit patterns of the concentric circular repeating pattern 22.

Next, the operation of the object appearance inspection apparatus of the present invention shown in FIG. 1 when inspecting the circular object 21 shown in FIG.

Prior to the inspection, it is necessary to set data for each model, reference position, reference rotation position, and each area. First, this data setting procedure will be described.

The operator operates the operation unit 3 to select the data setting mode as the operation mode.

FIG. 3 is a flow chart showing the processing procedure in the data setting mode of the object appearance inspection apparatus of the present invention.

In step 31 of FIG. 3, the operator operates the operation section 3 to operate the image pickup apparatus 1 via the control section 11 to pick up an image of a reference object as a comparison reference.

The control unit 11 temporarily stores the reference image obtained by picking up the image of the reference object in the image storage unit 2 and causes the display unit 10 to display the reference image.

Next, in step 32, the operator operates the operation unit 3 while looking at the reference image displayed on the display unit 10, and the misregistration correction model including the peripheral portion of the reference object is displayed on the reference image. A rotation deviation correction model area including any one of the area and the unit pattern of the reference object is set.

The above area setting is performed by inputting coordinate data of each area from the keyboard or by clicking each area on the displayed reference image with a mouse.

The control unit 11 temporarily stores in the data storage unit 6 the two vertex coordinates on the diagonal line of the set positional deviation correction model area and rotational deviation correction model area.

One or a plurality of misregistration correction model regions are set according to the desired misregistration correction accuracy.

Here, it is assumed that two misregistration correction model areas are set, an area including the left or right peripheral portion of the reference object image is set as the misregistration correction model area, and An area including the upper or lower peripheral portion is set as the second misregistration correction model area.

When the number of misregistration correction model areas is two as described above, the area including the left or right peripheral portion and the area including the upper or lower peripheral portion of the reference object image should be set. Is desirable in terms of positional deviation correction accuracy.

FIG. 4 shows an example of setting the above-mentioned misregistration correction model area and rotation deviation correction model area.

In FIG. 4, 41 is a reference image, 42 is an image of a reference object, and 43 is an image of a unit pattern on the reference object.

Reference numeral 44 shows a setting example of the first misregistration correction model area, 45 shows a setting example of the second misregistration correction model area, and 46 shows a setting example of the rotation deviation correction model area. Is shown.

Returning to FIG. 3, in step 33, the control unit 1
1 operates the model creating unit 4 to create a positional deviation correction model and a rotational deviation correction model.

That is, the model creating section 4 creates the first and second misregistration correction models by cutting out the images of the first and second misregistration correction model areas from the reference image, and rotates them. A rotation deviation correction model is created by cutting out an image in the deviation correction model area, and the position deviation correction model and the rotation deviation correction model are stored in the image storage unit 2.

Subsequently, at step 34, the control unit 11
Operates the reference value calculation unit 5 to calculate the reference position and the reference rotational position.

That is, the reference value calculation section 5 sets the intersection point coordinates of the diagonal lines in each area as the center point coordinates of the area,
If the center point coordinates of the first misregistration correction model area are (X ₁ , Y ₁ ) and the center point coordinates of the second misregistration correction model area are (X ₂ , Y ₂ ), a straight line x = X ₁ and y
= Y ₂ intersection point coordinates are obtained as the eigenpoint coordinates of the reference object,
This unique point coordinate is set as a reference position (X ₀ , Y ₀ ).

In the above case, the reference position (X ₀ , Y ₀ ) =
(X ₁ , Y ₂ ).

If the center point coordinates of the rotational displacement correction model area are (X ₃ , Y ₃ ), the center point coordinates (X ₃ ,
Y ₃ ) and a reference position (X ₀ , Y ₀ ) and a straight line y = Y ₀ form an angle formed by the reference object as a rotation angle of the reference object, and this rotation angle is set as a reference rotation position θ ₀ . The calculated reference position and reference rotational position are stored in the data storage unit 6.

As the reference position, the midpoint coordinates of a line segment whose end points are the central point coordinates of the two misregistration correction model areas may be used.

FIG. 5 illustrates the calculation of the reference position and the reference rotational position.

In FIG. 5, reference numeral 41 shows a reference image, and 42 shows an image of the reference object.

Reference numeral 44 denotes a first misregistration correction model area, 45 denotes a second misregistration correction model area, and 46 denotes a rotation deviation correction model area.

In the first misregistration correction model area 44
The center point coordinates (diagonal intersection coordinates) are (X₁, Y₁),second
The center point coordinates of the misregistration correction model area 45 are
(X_Two, Y _Two), And the straight line x = X₁And y = Y_TwoIntersection of
Coordinates (X₁, Y_Two) Is the eigenpoint coordinate of the reference object,
The reference position (X₀, Y₀).

Further, the center point coordinates of the rotation deviation correction model area 46 are (X ₃ , Y ₃ ), and two points of the center point coordinates (X ₃ , Y ₃ ) and the reference position (X ₀ , Y ₀ ) are set. The angle θ ₀ formed by the straight line passing through and the straight line y = Y ₀ is the rotation angle of the reference object, and is therefore the reference rotation position.

Next, in step 35, the operator operates the operation part 3 while looking at the reference image displayed on the display part 10 to include the rotation deviation correction model area set in step 32 and A search area including two adjacent side halves of two unit patterns adjacent to the correction model area is set on the reference image, and the control unit 11 sets the two vertex coordinates on the diagonal of the set search area to the data storage unit. 6
To memorize.

Next, in step 36, the operator operates the operation unit 3 while looking at the reference image displayed on the display unit 10 to set at least one inspection area including all or a part of the reference object. The control unit 11 causes the data storage unit 6 to store the two vertex coordinates on the diagonal line of the set inspection area region.

FIG. 6 is a diagram showing an example of setting the search area in step 35 of FIG. 7 and an example of setting the inspection area in step 36 of FIG.

In FIG. 6, reference numeral 41 indicates a reference image, and 4
Reference numeral 2 is an image of the reference object, and 43 is an image of a unit pattern on the reference object.

Reference numeral 46 shows a setting example of the rotational displacement correction model area, and 61 shows a setting example of the search area.

Reference numeral 62 shows an example of setting the inspection area. This inspection area 62 is an example of the setting of the inspection area when only the concentric circular repeating pattern on the circular object is the object of the appearance inspection.

Reference numeral 63 shows an example of setting the inspection area when the entire circular object is the object of the appearance inspection.

Here, it is assumed that the inspection area 62 is set.

Returning to FIG. 3, in step 37, the control section 11 operates the model creating section 4 to create an inspection model.

That is, the model creating section 4 creates an inspection model by cutting out images of the respective inspection areas from the reference image, and stores each of the inspection models in the image storage section 2.

With that, the processing flow in the data setting mode ends.

Next, the inspection execution procedure will be described.

The operator operates the operation unit 3 to select the inspection execution mode and input the inspection execution command.

FIG. 7 is a flow chart showing the processing procedure in the inspection execution mode of the object appearance inspection apparatus of the present invention.

In step 71 of FIG. 7, the control unit 11
Operates the image pickup apparatus 1 to capture an image of the inspection target object to obtain an inspection target image, and temporarily stores the inspection target image in the image storage unit 2.

Next, at step 72, the control section 11
The misregistration correction unit 7 is operated to extract regions (pattern misalignment correction model corresponding regions) that pattern-match with each misregistration correction model in the inspection target image.

FIG. 8 is a diagram for explaining the extraction of the misalignment correction model corresponding area.

In FIG. 8, reference numeral 81 is a misregistration correction model, and 82 is an image to be inspected.

The misregistration correction model 81 is assumed to be an image of N × N pixels, and the pixel at the upper left of the image has coordinate values (0,
0), and the lower right pixel of the image is represented by coordinate values (N-1, N-
1).

The image 82 to be inspected is assumed to be an image of M × M pixels (M> N), the upper left pixel of the image is represented by the coordinate value (0, 0), and the lower right pixel of the image is represented by the coordinate value ( M-1,
M-1).

The extraction of the misalignment correction model corresponding region is performed by
An area of N × N pixels is set in the inspection target image 82, and the image of each set area and the position shift correction model 8 are set.
The correlation value with 1 is calculated, and the region having the highest correlation value is set as the corresponding region.

83 is the coordinates (a, b) and (N + a-).
1, N + b-1) is shown as an area of N × N pixels.

Since 0 ≦ a ≦ M−N and 0 ≦ b ≦ M−N, the number of N × N pixel regions in the image 82 to be inspected is (M−N + 1) × (M−N + 1). Become.

The correlation value between the image of the area having the upper left apex coordinates of (a, b) and the misregistration correction model 81 is C (a,
b), the correlation value C (a, b) is calculated by the following formula.

[0093]

(Equation 1) In the above (Equation 1), T (x, y) is the brightness level of the pixel at the coordinates (x, y) of the misregistration correction model 81, I
(X + a, y + b) is the luminance level of the pixel of the coordinates (x + a, y + b) in the N × N pixel area set on the inspection target image 82.

Returning to FIG. 7, in step 73, the control unit 1
Reference numeral 1 operates the displacement correction unit 7 to calculate the eigenpoint coordinates of the inspection target object based on the positional relationship between the extracted misalignment correction model corresponding regions, and the eigenpoint coordinates and the reference position described above. The amount of displacement of the object to be inspected is calculated from the difference between

The eigenpoint coordinates of the object to be inspected are calculated based on the positional relationship of the extracted misalignment correction model corresponding regions.
It is calculated in the same manner as the calculation of the reference position shown in step 34 of FIG.

That is, the center point coordinates (intersection coordinates of diagonal lines) of the extracted first positional deviation correction model corresponding area are set to (X
₁ ′, Y ₁ ′) and the center point coordinates of the second misregistration correction model corresponding region are (X ₂ ′, Y ₂ ′), a straight line x =
The coordinates of the intersection of X ₁ ′ and y = Y ₂ ′ are obtained, and these are set as the eigenpoint coordinates (X ₀ ′, Y ₀ ′) of the inspection object. In this case, the eigenpoint coordinates (X ₀ ′, Y ₀ ′) = (X
₁ and Y ₂ ′).

Then, the eigenpoint coordinates (X
₀ ', Y _0' determined by) the difference between the reference position _{(X 0, Y 0) dX} , following equation _{dY, dX = X 0 -X 0} 'dY = Y 0 -Y 0', this dX , DY are the positional shift amounts of the inspection object.

Subsequently, at step 74, the control unit 11
Activates the position shift correction unit 7 to set the inspection target image to X
Scroll dX in the direction and dY in the Y direction (translation)
By doing so, the inspection target image whose positional displacement amount is corrected is obtained, and the inspection target image whose positional displacement is corrected is stored in the image storage unit 2.
To be temporarily stored in.

FIG. 9 illustrates the calculation of the positional deviation amount in step 73 of FIG. 7 and the positional deviation correction in step 74.

In FIG. 9, reference numeral 91 indicates an inspection target image (before positional deviation correction), and reference numeral 92 indicates an inspection target image whose positional deviation is corrected.

Reference numeral 93 indicates an image of the object to be inspected.
Reference numeral 4 denotes a first misregistration correction model corresponding area, and 95 denotes a second misregistration correction model corresponding area.

In the inspection target image 91, the center point coordinates of the first displacement correction model corresponding region 94 are (X ₁ ′,
Y ₁ ′), the center point coordinates of the second misregistration correction model corresponding area 95 are (X ₂ ′, Y ₂ ′), and the eigenpoint coordinates of the inspection object are (X ₀ ′, Y ₀ ′). (= (X ₁ ′, Y ₂
´)).

Then, the inspection target image 9 whose position has been corrected
2, the image 93 of the inspection target object is translated in the X direction by the displacement amount dX and the Y direction by the displacement amount dY, and the eigenpoint of the inspection target object is the reference position (X ₀ , Y ₀ ).
Is mapped to

Returning to FIG. 7, in step 75, the control unit 1
Reference numeral 1 operates the rotation deviation correction unit 8 to extract an area (rotation deviation correction model corresponding area) that is pattern-matched with the rotation deviation correction model in the search area on the inspection target image whose position deviation has been corrected. To do.

The extraction of the rotation misalignment correction model corresponding area is performed by the same method as the extraction of the misregistration correction model corresponding area in step 72.

Subsequently, at step 76, the control unit 11
Operates the rotation deviation correction unit 8 to calculate the rotation angle of the inspection target object based on the positional relationship between the extracted rotation deviation correction model corresponding region and the reference position, and calculates the rotation angle between the rotation angle and the reference rotation position. The amount of rotation deviation of the object to be inspected is calculated from the difference.

That is, the central point coordinates (intersection coordinates of diagonal lines) of the extracted rotational displacement correction model corresponding area are set to (X ₃ ′,
Y ₃ ′), a straight line passing through the two points of the center point coordinates (X ₃ , Y ₃ ) and the reference position (X ₀ , Y ₀ ),
The angle formed by the straight line y = Y ₀ is the rotation angle θ ₀ ′ of the inspection object.
Then, the difference θ between the rotation angle θ ₀ ′ of the object to be inspected and the reference rotational position θ ₀ is obtained by the following equation, θ = θ ₀ −θ ₀ ′, and this θ is the amount of rotational deviation of the object to be inspected. And

Subsequently, in step 77, the control unit 11
Causes the rotation deviation correction unit 8 to operate, and performs an affine transformation process for rotating the inspection target image whose position deviation has been corrected by θ around the reference position (X ₀ , Y ₀ ). The inspection target image with the corrected amount and the rotational displacement amount is obtained, and the inspection target image with the corrected positional displacement and rotational displacement is temporarily stored in the image storage unit 2.

The coordinates of an arbitrary pixel in the inspection target image (inspection target image before affine transformation processing) whose position has been corrected are defined as (X, Y), and the pixel at this coordinate (X, Y) is subjected to affine transformation processing. When the coordinates are (U, V), the above affine transformation process is expressed by the following equation.

[0110]

(Equation 2) FIG. 10 illustrates the calculation of the rotation deviation amount in step 76 of FIG. 7 and the rotation deviation correction in step 77.

In FIG. 10, reference numeral 101 denotes an inspection object image whose position deviation has been corrected (before rotation deviation correction), and 102 denotes an inspection object image whose position deviation and rotation deviation have been corrected.

Further, 103 is an image of the object to be inspected, 104
Shows the image of the unit pattern of the inspection object,
Reference numeral 105 denotes a rotation misalignment correction model corresponding area.

In the inspection target image 101 whose position has been corrected, the center point coordinates of the rotation deviation correction model corresponding area 105 are (X ₃ ′, Y ₃ ′), and the center point coordinates (X
₃ , Y ₃ ) and the reference position (X ₀ , Y ₀ ) and the straight line y = Y ₀ forms the angle of rotation of the inspection object, which is θ ₀ ′.

Then, in the inspection target image 102 whose position and rotation are corrected, the image 1 of the inspection target object is obtained.
No. 03 is rotated about the reference position (X ₀ , Y ₀ ) by the rotation deviation amount θ, the rotation angle of the inspection target object is θ ₀ , and the inspection target object is mapped to the reference rotation position θ ₀ . .

Returning to FIG. 7, in step 78, the control unit 1
1 is a defect in the appearance of the inspection target object by operating the inspection execution unit 9 and comparing the image of the inspection region in the inspection target image corrected for the positional deviation and the rotational deviation with the inspection model. Inspect for the presence of

An example of the appearance inspection of the inspection object will be described with reference to FIG.

FIG. 11 shows the case where only the concentric repeating pattern of a circular object is the object of inspection.

In FIG. 11, 111 indicates each of the first to sixth inspection models, 112 indicates the inspection target image in which the positional deviation and the rotational deviation are corrected, and 113 is set corresponding to each inspection model 111. The respective first to sixth inspection areas are shown.

It is assumed that the inspection model 111 and the inspection area 113 have the same angle of view and are N ′ × N ′ pixel images.

The upper left vertex coordinates of each inspection model 111 are (0, 0), and the lower right vertex pixels are (N'-1, N'-.
1).

The coordinates of the upper left corner of the image in each inspection area 113 are set to (A _k , B _k ) and the coordinates of the lower right corner of the image are set to ( _{N ′} ).
+ A _k -1, N '+ B _k -1). Here, k is a variable indicating the number of the inspection area, and takes any one of 1, 2, ...

The appearance inspection of the inspection object is performed by calculating the correlation value between the images of the corresponding inspection model 111 and the inspection area 113, like the images of the inspection model 111-1 and the inspection area 113-1. It is to determine whether or not the value is equal to or greater than a preset reference value, and if the calculated correlation value is equal to or greater than the reference value, the concentric circular pattern displayed in the inspection area is displayed. Is judged to have no visual defects.

When the correlation value between the kth inspection model and the image of the kth inspection region is C (k) (k = 1, 2 ... 6), the correlation value C (k) is given by (Equation 1). Similarly, it is calculated by the following formula.

[0124]

(Equation 3) In the above (Equation 3), T _k (x, y) is the brightness level of the pixel at the coordinates (x, y) of the kth inspection model, and I _k (x
+ A _k , y + B _k ) is the coordinate (x
+ A _k , y + B _k ) pixel luminance level.

As described above, according to the above-described embodiment, at least one position deviation correction area including the peripheral portion of the reference object and the rotation deviation correction area including an arbitrary unit pattern are provided on the reference image by the operation unit 3. By setting an area, a search area including this rotation deviation correction area and including adjacent halves of two unit patterns adjacent to this area, and at least one inspection area including all or a part of the reference object, The search area and the inspection area are stored in the data storage unit 6 in advance, the reference value calculation unit 5 calculates the reference position of the object based on the position of the position deviation correction area, and the rotation deviation correction is performed. The reference rotational position of the object is calculated based on the positional relationship between the working area and the reference position, and the reference position and the reference rotational position are stored in advance in the data storage unit 6, The image forming unit 4 cuts out the image of the misregistration correction area, the image of the rotational misregistration correction area, and the image of the inspection area from the reference image to obtain the misregistration correction model and the rotational deviation correction image. A model and an inspection model are created, each created model is stored in advance in the image storage unit 2, and the position deviation correction unit 7 extracts an area corresponding to the position deviation correction model in the inspection target image. Then, the eigenpoint coordinates of the object to be inspected are calculated based on the position of the extracted area, the amount of displacement of the object to be inspected is calculated from the difference between the coordinates of the eigenpoint and the reference position, and the predetermined value is set with respect to the image to be inspected. By performing the scroll processing of, the position deviation amount is corrected,
An area corresponding to the rotational deviation correction model is extracted from the search area in the inspection target image whose positional deviation has been corrected by the rotational deviation correction unit 8, and based on the positional relationship between the extracted area and the reference position. The rotation angle of the inspection target object is calculated, the rotation displacement amount of the inspection target object is obtained from the difference between this rotation angle and the reference rotation position, and a predetermined affine transformation process is performed on the position displacement corrected inspection target image. By performing the correction of the rotation deviation amount by performing, by comparing the image of the inspection region in the inspection target image in which the position deviation and the rotation deviation have been corrected by the inspection execution unit 9 and the inspection model, By inspecting the appearance of the object to be inspected for defects, circular objects with concentric circular repeating patterns are placed at unspecified positions and rotation angles. By using an image that, the presence or absence of a defect in appearance of the circular object can be a short time to inspect with high accuracy.

[0126]

As described above, according to the image processing method of the first aspect and the image processing apparatus of the third aspect of the present invention, based on the position of the misregistration correction area set by the operating means. For calculating the reference position of the object by the reference value calculating means, storing the reference position in the reference value storing means, and for correcting the position deviation by cutting out the image of the position deviation correcting area from the reference image by the model creating means. A model is created and stored in advance in the model storage means, and the reference rotation position of the object is determined by the reference value calculation means based on the positional relationship between the rotation deviation correction area set by the operation means and the reference position. The calculated reference rotation position is stored in the reference value storage means, and the image of the rotation deviation correction area is cut from the reference image by the model creation means. A rotation deviation correction model is created by storing it in advance in the model storage means, and the position deviation amount calculation means extracts and extracts a region corresponding to the position deviation correction model in the processing target image. The eigenpoint coordinates of the object to be processed are calculated based on the position of the area, and the amount of positional deviation of the object to be processed is calculated from the difference between the eigenpoint coordinates of the object to be processed and the reference position, and the object to be processed by the positional deviation correcting means. By performing a predetermined scrolling process on the image, the positional deviation amount is corrected, and the rotational deviation amount calculation means extracts a region corresponding to the rotational deviation correction model from the positional deviation corrected processing target image. Then, the rotation angle of the processing target object is calculated based on the positional relationship between the extracted region and the reference position, and the rotation angle of the processing target object and the reference rotation position are calculated. The rotational deviation amount of the object to be processed is calculated from the difference between the two, and the processing target image subjected to the positional deviation correction by the rotational deviation correction means is subjected to a predetermined affine transformation process to correct the rotational deviation amount, thereby concentric circles. In an image of a circular object having a circular repeating pattern, the circular object whose position and rotation angle are unspecified can be accurately arranged at a predetermined position and a predetermined rotation position.

According to the image processing method according to claim 2 and the image processing device according to claim 4, the image processing method according to claim 1 and the image processing device according to claim 3 of the present invention are: A search area, which is set on the reference image by the operation means, includes the rotation deviation correction area and includes two adjacent side halves of two unit patterns adjacent to this area, is stored in advance in the area storage means, and is rotated. A circular object having a concentric circular repeating pattern is determined in a short time by extracting a region corresponding to the rotational displacement correction model in the search region in the processing target image whose positional displacement amount has been corrected by the displacement amount calculating means. There is an effect that it can be arranged at a predetermined position and a predetermined rotation position with high accuracy.

Next, according to the object appearance inspection apparatus of the fifth aspect of the present invention, the operation means sets the position deviation correction area, the rotation deviation correction area, and the inspection area on the reference image, The inspection area is stored in advance in the area storage means, the reference value calculation means calculates the reference position of the object based on the position of the position deviation correction area, and the rotation deviation correction area and the reference position are calculated. The reference rotational position of the object is calculated based on the positional relationship of the object, the calculated reference position and the reference rotational position are stored in the reference value storage means in advance, and the model creation means uses the reference image to detect the position deviation correction area. Image, the image of the rotation deviation correction area, and the image of the inspection area are respectively cut out to obtain a position deviation correction model, a rotation deviation correction model, and an inspection model. The created model is stored in advance in the model storage means, and the position shift amount calculation means extracts a region corresponding to the position shift correction model in the inspection target image, and the position of the extracted region is set. The eigenpoint coordinates of the object to be inspected are calculated based on the difference between the eigenpoint coordinates and the reference position, and the amount of positional deviation of the object to be inspected is determined. By correcting the positional deviation amount, the rotational deviation amount calculating means extracts an area corresponding to the rotational deviation correction model in the inspection target image whose positional deviation is corrected, and extracts the extracted area and the extracted area. The rotation angle of the inspection target object is calculated based on the positional relationship with the reference position, and the rotation deviation amount of the inspection target object is obtained from the difference between this rotation angle and the reference rotation position, The rotation deviation amount is corrected by performing a predetermined affine transformation process on the inspection target image whose position deviation has been corrected by the rotation deviation correction means, and the position deviation and the rotation deviation have been corrected by the inspection execution means. By comparing the image of the inspection area in the inspection target image and the inspection model, by inspecting the presence or absence of a defect on the appearance of the inspection target object, the circular object having a concentric circular repeating pattern is unspecified. There is an effect that the presence or absence of a defect in the appearance of the circular object can be inspected with high accuracy by using the images arranged at the position and the rotation angle.

According to a sixth aspect of the present invention, there is provided the object visual inspection apparatus according to the fifth aspect, which includes the rotation deviation correction area set on the reference image by the operating means. Further, a search area including two adjacent side halves of two unit patterns adjacent to this area is stored in advance in the area storage means, and the search area in the processing target image whose position deviation amount is corrected by the rotation deviation amount calculation means. In, by extracting a region corresponding to the rotation deviation correction model, by using an image in which the circular object is arranged at an unspecified position and a rotation angle, on the appearance of the circular object having a concentric repeating pattern There is an effect that the presence or absence of defects can be inspected with high accuracy in a short time.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of an embodiment of an object appearance inspection device of the present invention.

FIG. 2 is an external view showing an example of a circular object having a concentric circular repeating pattern, which is inspected by an object appearance inspection device.

FIG. 3 is a flowchart showing a processing procedure in a data setting mode of the object appearance inspection device of the present invention.

FIG. 4 is a diagram showing a setting example of a positional deviation correction model area and a rotational deviation correction model area in step 32 of FIG. 3;

5 is an explanatory diagram showing calculation of a reference position and a reference rotational position in step 34 of FIG.

6 is a diagram showing an example of setting a search area in step 35 and an example of setting an inspection area in step 36 of FIG.

FIG. 7 is a flowchart showing a processing procedure in an inspection execution mode of the object appearance inspection apparatus of the present invention.

8 is an explanatory diagram showing extraction of a misregistration correction model corresponding region in step 72 of FIG. 7. FIG.

9 is an explanatory diagram showing the calculation of the positional deviation amount in step 73 and the positional deviation correction in step 74 of FIG. 7. FIG.

10 is an explanatory diagram showing the calculation of the rotation deviation amount in step 76 and the rotation deviation correction in step 77 of FIG.

FIG. 11 is an explanatory diagram showing an example of the appearance inspection of the inspection target object in step 78, returning to FIG. 7.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Imaging device 2 Image storage unit 3 Operation unit 4 Model creation unit 5 Reference value calculation unit 6 Data storage unit 7 Positional deviation correction unit 8 Rotational deviation correction unit 9 Inspection execution unit 10 Display unit 11 Control unit

Claims (6)

[Claims] 1. Regarding a circular object in which a plurality of characteristic unit patterns are arranged concentrically, a reference position and a reference rotational position of the object are calculated in advance based on a reference image obtained by imaging the reference object. It is stored, in the processing target image obtained by imaging the processing target object, an image processing method of arranging the processing target object at the reference position and the reference rotation position, which is set on the reference image, The eigenpoint coordinates of the reference object are calculated based on the position of at least one misregistration correction region including the peripheral portion of the reference object, and the eigenpoint coordinates are stored as the reference position, and the misalignment correction is performed. An image of a region is cut out from the reference image, stored in advance as a position deviation correction model, and a rotation deviation including an arbitrary unit pattern set on the reference image is set. The rotation angle of the reference object is calculated based on the positional relationship between the normal use area and the reference position, the rotation angle is stored as the reference rotation position, and the image of the rotation deviation correction area is changed from the reference image. Cutting out, stored in advance as a rotation deviation correction model, and in the processing target image, an area corresponding to the position deviation correction model is extracted, and eigenpoint coordinates of the processing target object based on the position of the extracted area. Is calculated, the amount of positional deviation of the processing target object is obtained from the difference between the eigenpoint coordinates of the processing target object and the reference position, and the positional deviation is performed by performing a predetermined scrolling process on the processing target image. In the processing target image whose position has been corrected, the region corresponding to the rotation displacement correction model is extracted, and the position between the extracted region and the reference position is extracted. The rotation angle of the object to be processed is calculated based on the relationship, the amount of rotational deviation of the object to be processed is calculated from the difference between the rotational angle of the object to be processed and the reference rotational position, and the processing target image corrected for the positional deviation An image processing method, wherein the rotation deviation amount is corrected by performing a predetermined affine transformation process on the. 2. A search area set on the reference image is stored in advance, the search area including the rotation deviation correction area and two adjacent side halves of two unit patterns adjacent to the area. The image processing method according to claim 1, wherein an area corresponding to the rotation deviation correction model is extracted from the search area in the processing target image whose deviation amount has been corrected. 3. Regarding a circular object in which a plurality of characteristic unit patterns are concentrically arranged, a reference position and a reference rotational position of the object are calculated in advance based on a reference image obtained by imaging the reference object. An image processing apparatus that stores the processing target object in a processing target image obtained by capturing an image of the processing target object, and arranges the processing target object at the reference position and the reference rotation position, wherein the reference image is displayed on the reference image. Operation means for setting at least one position deviation correction area including the peripheral portion of the object, and for setting a rotation deviation correction area including any of the unit patterns, and based on the position of the position deviation correction area Calculate the eigenpoint coordinates of the reference object as the reference position, and determine the rotation angle of the reference object based on the positional relationship between the rotation deviation correction area and the reference position. A reference value calculation unit that calculates the position, and a position deviation correction model is created by cutting out the image of the position deviation correction area from the reference image, and rotation is performed by cutting out the image of the rotation deviation correction area. Model creating means for creating a deviation correction model, model storage means for storing the position deviation correction model and rotation deviation correction model, reference value storage means for storing the reference position and the reference rotation position, and the processing In the target image, each region corresponding to the displacement correction model is extracted, the eigenpoint coordinates of the processing target object are calculated based on the position of the extracted region, and the difference between the eigenpoint coordinates and the reference position is calculated. A positional shift amount calculating means for obtaining a positional shift amount of the processing target object from the above, and performing a predetermined scrolling process on the processing target image With this, a positional deviation correction unit that corrects the positional deviation amount, and an area corresponding to the rotational deviation correction model in the positional deviation corrected processing target image is extracted, and the extracted area and the reference position A rotation deviation amount calculation unit that calculates a rotation angle of the processing target object based on a positional relationship, and obtains a rotation deviation amount of the processing target object from a difference between the rotation angle and the reference rotation position; An image processing apparatus, comprising: a rotation deviation correction unit that corrects a rotation deviation amount of the processing object by performing a predetermined affine transformation process on the processing object image. 4. The operation means for setting a search area on the reference image, the search area including the rotation deviation correction area, and each half of the two adjacent unit patterns adjacent to the rotation deviation correction area. An area storage unit that stores the search area, and a rotation deviation amount calculation unit that extracts an area corresponding to the rotation deviation correction model in the search area in the position deviation corrected image to be processed. The image processing apparatus according to claim 3, wherein: 5. With respect to a circular object in which a plurality of characteristic unit patterns are concentrically arranged, an image of the object to be inspected is compared with an image of a reference object stored in advance. An object appearance inspection device for inspecting the presence or absence of a defect in appearance, wherein at least one position deviation correction area including a peripheral portion of the reference object is set on a reference image obtained by imaging the reference object. And an operation unit for setting a rotation deviation correction area including any of the unit patterns, and further setting at least one inspection area including all or part of the reference object, and the position deviation correction area. The eigenpoint coordinates of the reference object are calculated as the reference position based on the position of the reference object, and the rotation angle of the reference object is calculated based on the positional relationship between the rotation deviation correction area and the reference position. A reference value calculation unit for calculating a rotational position, and a positional deviation correction model is created by cutting out an image of the positional deviation correction area from the reference image, and an image of the rotational deviation correction area is cut out. A model creating unit that creates a rotation deviation correction model and further creates an inspection model by cutting out the image of the inspection area, an area storage unit that stores the inspection area, the position deviation correction model, and the rotation deviation. A model storage unit that stores a correction model and an inspection model, a reference value storage unit that stores the reference position and the reference rotation position, an inspection target image obtained by imaging the inspection target object, and the position A region corresponding to the displacement correction model is extracted, and the eigenpoint coordinates of the inspection target object are calculated based on the position of the extracted region, A positional shift amount calculating unit that obtains a positional shift amount of the inspection target object from the difference between the eigenpoint coordinates and the reference position, and a predetermined scroll process for the inspection target image to determine the positional shift amount. A misregistration correction unit that corrects, and an area corresponding to the rotational misalignment correction model in the inspection object image that has been misaligned, and the inspection object based on the positional relationship between the extracted area and the reference position. A rotation deviation amount calculating unit that calculates a rotation angle of the object and obtains a rotation deviation amount of the inspection object from the difference between the rotation angle and the reference rotation position, and a predetermined value for the position deviation corrected inspection object image. Deviation correction means for correcting the rotation deviation amount by performing the affine transformation processing of the above, and the inspection in the inspection target image in which the position deviation and the rotation deviation are corrected. An object appearance inspection apparatus comprising: an inspection execution means for inspecting the presence or absence of a defect on the appearance of the inspection object by comparing an image of a region with the inspection model. 6. The operation means for setting a search area on the reference image, the search area including the rotation deviation correction area, and each half of an adjacent side of two unit patterns adjacent to the rotation deviation correction area. An area storage unit that stores the search area; and a rotation deviation amount calculation unit that extracts an area corresponding to the rotation deviation correction model in the search area in the inspection target image whose position deviation has been corrected. The object appearance inspection apparatus according to claim 5, further comprising: