Probe mark inspection method and apparatus
View Patent ↗A method and apparatus for inspection of probe marks made on the interconnection lands of semiconductor devices using machine vision is disclosed. An image of an interconnection land is analyzed, and features of the image that may constitute indicia of probe marks are refined through the application of a series of unique heuristic processes. The output of the method is measurement data that can be used to characterize and verify the processes used to electrically probe semiconductor devices.
1. An image data analysis method for inspecting pads of electronic devices, the method comprising:
acquiring an image corresponding to a pad;
generating binarized image data from the image;
generating a list of dark regions from the binarized image;
selecting at least one dark region from the list of dark regions;
generating edge data by performing edge detection on a portion of the image corresponding to the at least one dark region;
positioning a geometric mask at a plurality of mask positions tangent to the edge data;
pruning the at least one dark region of at least a portion of the binarized image data not contained by the mask position; and
computing a boundary description of the at least one dark region using the binarized image data and the edge data.
2. The image data analysis method of claim 1 wherein the geometric mask is circular.
3. The image data analysis method of claim 1 further comprising the application of a fuzzy threshold according to an aspect ratio of the at least one dark region for limiting the pruning step.
4. An image data analysis method for inspecting pads of electronic devices, the method comprising:
acquiring an image corresponding to a pad;
generating binarized image data from the image;
generating a list of dark regions from the binarized image;
selecting at least one dark region from the list of dark regions;
generating edge data by performing edge detection on a portion of the image corresponding to the at least one dark region;
generating a first elliptical representation of an extremal end of the at least one dark region;
generating a second elliptical representation of a subset of the edge data corresponding to a region adjacent to both sides of the extremal end of the at least one dark region;
generating a third elliptical representation of the edges used in the first and second elliptical representations;
pruning the at least one dark region in response to a comparison of the first elliptical representation, the second elliptical representation and the third elliptical representation; and
computing a boundary description of the at least one dark region using the binarized image data and the edge data.
5. The image data analysis method of claim 4 wherein the step of pruning the at least one dark region further comprises:
calculating a fit error of the first elliptical representation and a fit error of the second elliptical representation and a fit error of the third elliptical representation; and
pruning the at least one dark region using the elliptical representation having a minimum fit error.
6. An image data analysis system comprising:
a camera;
a machine vision processor coupled to the camera, the processor and the camera configured to acquire an image corresponding to a region of the inspection;
means for generating binarized image data from the image;
means for generating a list of dark regions from the binarized image data;
means for selecting at least one dark region from the list of dark regions;
means for generating edge data by performing edge detection on a portion of the image corresponding to the at least one dark region;
means for heuristic refinement of the binarized image data and the edge data; and
means for computing a boundary description of the at least one dark region using the binarized image data and the edge data;
means for positioning a geometric mask at a plurality of mask positions, the mask positions tangent to the edge data in the image; and
means for pruning the at least one dark region of at least a portion of the boundary description not contained by the mask positions.
7. The system of claim 6 further comprising a means for applying a fuzzy threshold according to an aspect ratio of the at least one dark region for limiting the pruning means.
8. An image data analysis system comprising:
means for acquiring an image corresponding to a region of inspection;
means for generating binarized image data from the image;
means for generating a list of dark regions from the binarized image data;
means for selecting at least one dark region from the list of dark regions;
means for generating edge data by performing edge detection on a portion of the image corresponding to the at least one dark region;
means for generating a first elliptical representation of an extremal end of the at least one dark region;
means for generating a second elliptical representation of a subset of the edge data corresponding to a region adjacent to both sides of the extremal end of the at least one dark region;
means for generating a third elliptical representation of the edges used in the first and second elliptical representations; and
means for pruning the at least one dark region in response to a comparison of the first elliptical representation, the second elliptical representation and the third elliptical representation; and
means for computing a boundary description of the at least one dark region using the binarized image data and the edge data.
9. The system of claim 8 further comprising:
means for calculating a fit error of the first elliptical representation and a fit error of the second elliptical representation and a fit error of the third elliptical representation, and;
means for pruning the at least one dark region using the elliptical representation having a minimum fit error.
10. An image data analysis method for inspecting scenes, the method comprising:
acquiring an image of a scene;
generating binarized image data from the image;
generating a list of dark regions from the binarized image;
selecting at least one dark region from the list of dark regions;
generating edge data by performing edge detection on a portion of the image corresponding to the at least one dark region;
positioning a geometric mask at a plurality of mask positions tangent to the edge data;
pruning the at least one dark region of at least a portion of the binarized image data not contained by the mask positions; and
computing a boundary description of the at least one dark region using the binarized image data and the edge data.
11. The image data analysis method of claim 10 wherein the geometric mask is circular.
12. The image data analysis method of claim 10 further comprising the application of a fuzzy threshold according to an aspect ratio of the at least one dark region for limiting the pruning step.
13. An image data analysis method for inspecting scenes, the method comprising:
acquiring an image of a scene;
generating binarized image data from the image;
generating a list of dark regions from the binarized image;
selecting at least one dark region from the list of dark regions;
generating edge data by performing edge detection on a portion of the image corresponding to the at least one dark region;
generating a first elliptical representation of an extremal end of the at least one dark region;
generating a second elliptical representation of a subset of the edge data corresponding to a region adjacent to both sides of the extremal end of the at least one dark region;
generating a third elliptical representation using the edged from the first and second elliptical representations; and
pruning the at least one dark region in response to a comparison of the first elliptical representation, the second elliptical representation, and the third elliptical representation; and
computing a boundary description of the at least one dark region using the binarized image data and the edge data.
14. The image data analysis method of claim 13 wherein the step of pruning the at least one dark region further comprises:
calculating a fit error of the first elliptical representation and a fit error of the second elliptical representation and a fit error of the third elliptical representation; and
pruning the at least one dark region using the elliptical representation having a minimum fit error.