Automatic visual inspection apparatus, automatic visual inspection method, and recording medium having recorded an automatic visual inspection program

Claim 2

The automatic visual inspection apparatus according to claim 1, characterized in that said sub-pixel precision position displacement computing unit (6) computes the average value of the position displacement in sub-pixel precision in all edges in one divisional differential image as the position displacement of the sub-pixel precision of said divisional differential image.

Claim 3

The automatic visual inspection apparatus according to claim 1, characterized in that the image comparison unit (8) obtains the absolute value of the pixel value by the adjusted divisional differential image of the inspection image and the adjusted divisional differential image of the reference image, and in a case that the resulting value exceeds the predetermined threshold value, the unit regard said result to be the defect of the object to be inspected.

Claim 4

The automatic visual inspection apparatus according to claim 1, characterized in that said image comparison unit (8) forms the adjusted divisional differential image having the position displacement by a sub-pixel by interpolating the intensity between the pixels in either one of the adjusted divisional differential image, and compares it with the other adjusted differential image.

Claim 5

The automatic visual inspection apparatus according to claim 1, characterized in that the sub-pixel precision position displacement computing unit (6) detects a corner on the divisional differential image, and if a corner exists, the unit effect laying over the images in the pixel precision position alignment unit, after which, if, with respect to the certain corner on the inspected image, there is a corner in the region on the reference image corresponding to the nearest part to said corner, these two corners are regarded to be in corresponding relations, and the difference of the mutual corner positions is computed as the position displacement of the sub-pixel precision between the corners.

Claim 6

An automatic visual inspection method comprising the steps of: inputting an inspection image on the object to be inspected and a reference image for making comparison with the inspection image, and generating a differentiated images by spatially differentiating the respective inputted images; generating divisional differential images by dividing the differential images according to the predetermined number of divisions; overlaying the inspection image and the reference image by carrying out the position alignment in pixel precision which is a pixel size precision between the inspection image and the reference image on each formed divisional differential image; obtaining a coordinate of an edge, by regarding the point where the absolute pixel value is larger than the predetermined threshold value to be an edge in each divisional differential image, obtaining the direction of the edge, obtaining the intensity profile of the absolute pixel value in the direction perpendicular to the direction of the edge, fitting a predetermined single peak function to it, obtaining the maximum value point which is a coordinate of the point where its function takes the maximum value by a sub-pixel precision which is the precision lower than the pixel size and employing the resulting value as said coordinate of the edge; regarding, after the mutual laying over of the images, if, on certain edge on the inspection image, there is an edge in the region on the reference image corresponding to the neighborhood of the edge, these two edges to be in the corresponding relations, and computing the difference of the mutual edge positions as a position displacement of the sub-pixel precision; forming an adjusted differential image by rewriting the pixel value so that the intensity profile in the direction orthogonal with the edge should follow the previously given single peak type function form on the corresponding two edges; detecting the defect of the object to be inspected by comparing the adjusted divisional differential image of the inspected image with the adjusted divisional differential image of the reference image based on the position displacement of the sub-pixel precision of each divisional differential image computed.

Claim 7

An automatic visual inspection method comprising the steps of: inputting an inspection image on the object to be inspected and a reference image for comparison with said inspection image, and dividing the respective images according to the predetermined number of divisions; spatially differentiating the divided image to form a divisional differential image; overlaying an inspection image and a reference image by carrying out position alignment in pixel precision which is a pixel size precision between the inspection image and the reference image on each formed divisional differential image; obtaining a coordinate of an edge, by regarding the point where the absolute pixel value is larger than the predetermined threshold value to be an edge in each divisional differential image, obtaining the direction of the edge, obtaining the intensity profile of the absolute pixel value in the direction perpendicular to the edge, fitting a predetermined single peak function to it, obtaining the maximum value point which is a coordinate of the point where its function takes the maximum value by a sub-pixel precision which is the precision lower than the pixel size, and employing the resulting value as the coordinate of the edge; obtaining sub-pixel precision position displacement, by, after the laying over of the images together in the pixel precision position alignment step, if there is an edge in the region on the reference image corresponding to the neighborhood of the edge, these two edges are regarded as being in the corresponding relations, and computing the difference of the mutual edge positions as a position displacement of the sub-pixel precision; forming an adjusted differential image by rewriting the pixel value so that the intensity profile of the pixel value in the direction orthogonal with the edge should follow the previously given single peak type function form on the corresponding two edges; and comparing the adjusted divisional differential image of the inspected image with the adjusted divisional differential image of the reference image based on the position displacement of the sub-pixel precision of each divisional differential image computed as above.

Claim 8

A recording medium having recorded a program for causing a computer to execute: an image division differential processing having inputs of an inspection image on the object to be inspected and a reference image for making comparison with the inspection image, spatially differentiating the respective inputted images and dividing them into the predetermined number of divisions to form the divisional differential images; a pixel precision position alignment processing for laying the inspection image and the reference image together by carrying out the position alignment in pixel precision which is a pixel size precision between the inspection image and the reference image on each of the divisional differential image formed by the image division differential unit; a function fitting processing so designed as to regard the point where the absolute pixel value is larger than the predetermined threshold value to be an edge in each divisional differential image, obtaining the direction of the edge, obtaining the intensity profile of the absolute pixel value in the direction orthogonal with the edge, fitting a predetermined single peak function to it, obtaining the maximum value point which is a coordinate of the point where its function takes the maximum value by a sub-pixel precision which is the precision lower than the pixel size, and employing the resulting value as coordinate of an edge; a sub-pixel precision position displacement computing processing which functions so that, after the laying over of the images together in the pixel precision position alignment processing, if there is an edge in the region on the reference image corresponding to the neighborhood of the edge, these two edges are regarded as being in the corresponding relations, and computing the difference of the mutual edge positions as a position displacement of the sub-pixel precision; an adjusted differential image formation processing for forming the adjusted differential image by rewriting the pixel value so that the intensity profile of the pixel value in the direction orthogonal with the edge should follow the previously given single peak type function form on the corresponding two edges; and an image comparison processing for detecting the defect of the object to be inspected by comparing the adjusted divisional differential image of the inspected image with the adjusted divisional differential image of the reference image based on the position displacement of the sub-pixel precision of each divisional differential image computed by the sub-pixel precision position displacement computing unit.