Machine vision system for object feature analysis and validation based on multiple object images

• US 6,175,644 B1
• Filed: 05/01/1998
• Issued: 01/16/2001
• Est. Priority Date: 05/01/1998
• Status: Active Grant

• Current Assignee: Cognex Corporation, Russ Weinzimmer
• Original Assignee: Cognex Corporation
• Inventors: Scola, Joseph R., Ruzhitsky, Vladimir N., Jacobson, Lowell D.
• Primary Examiner(s): MEHTA, BHAVESH M
• Application Number:
  US09/071,722
• Time in Patent Office: 991 Days
• Field of Search: 382/107, 382/141, 382/143, 382/293, 382/190, 382/199, 382/266, 382/216, 348/154, 348/155, 117/14, 117/15, 117/201, 117/202
• US Class Current: 382/141
• CPC Class Codes: G01N 21/8851 Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges G01N21/8806 and G01N21/93 - G01N21/95692 take precedence; optical measurement of dimensions G01B11/00; optical scanning G02B26/10; image transformation G06T3/00; computerised image enhancement G06T5/00; image processing per se for flaw detection G06T7/0002
• CPC Class Codes: G06T 7/0004 Industrial image inspection
• CPC Class Codes: Y10T 117/1004 with means for measuring, testing, or sensing

Abstract

Provided is the ability to validate detected features in acquired images to thereby enhance the integrity of any analysis carried out on the detected and validated features. A sequence of images of, e.g., an object is acquired, each image in the sequence corresponding to a distinct orientation of the object about a selected object axis. Images in the sequence are inspected for feature points of the selected feature plane, as-projected into the images, at a first feature detection location and at a second feature detection location. The second feature detection location is configured at an image position at which a feature point detected in the first feature detection location in a first inspected image is expected to appear in a second inspected image. Valid object feature points are identified as being those feature points which are detected in both the first feature detection location in a first inspected image and in the second feature detection location in a second inspected image of the image sequence. Features that are validated are not likely to be time-dependent noise and are preserved for further feature analysis, while extraneous data is rendered transparent to feature analysis. Further, an object feature plane can be analyzed for a specified feature configuration even when only a subset of feature points is available in any one given image of the object, e.g., where a portion of a three-dimensional object obscures other portions of the object in a selected view of the object. Also provided is a method for detecting features of a semiconductor melt surface as a semiconductor ingot is pulled out of the melt.

Claims

1. A machine vision method for detecting features of an object in a selected feature plane of the object, the method comprising:

acquiring a sequence of images of the object from a fixed viewing location as the object rotates about a selected object axis, each image in the sequence corresponding to a distinct orientation of the object about the selected object axis, the sequence of acquired object images comprising at least a minimum number, M, of object images, where M is selected as M=T₀/T_a, where T₀ is a period of revolution of the object and T_a is a period required for acquiring one image;

inspecting images in the image sequence for feature points of the selected feature plane, as projected into the images, at a number, n, of feature detection locations, the feature detection locations configured with respect to each other at image positions based on an expected shift in image location of a given feature point between two consecutive images in the image sequence;

associating detected feature points with an orthogonal-view representation of the selected feature plane; and

identifying as valid object feature points those feature points detected in a specified minimum number of the n feature detection locations each in a different image of the image sequence.

2. A machine vision method for detecting features of an object in a selected feature plane of the object, the method comprising:

acquiring a sequence of images of the object, each image in the sequence corresponding to a distinct orientation of the object about a selected object axis;

inspecting images in the image sequence for feature points of the selected feature plane, as projected into the images, at a number, n, of feature detection locations, by;

applying an edge detection region at each of the n feature detection locations,

searching each edge detection region for an edge of an object feature point; and

correlating positions of detected edges to object feature point positions,

the feature detection locations configured with respect to each other at image positions based on an expected shift in image location of a given feature point between two consecutive images in the image sequence;

associating detected feature points with an orthogonal-view representation of the selected feature plane; and

identifying as valid object feature points those feature points detected in a specified minimum number of the n feature detection locations each in a different image of the image sequence.

3. A machine vision method for detecting features of a semiconductor melt surface from which a semiconductor crystal ingot is pulled as the ingot and melt are rotated about a vertical axis of rotation, the method comprising:

acquiring a sequence of perspective-view images of the melt surface from a fixed viewing location, each image in the sequence corresponding to a perspective view of a distinct orientation of the melt about the vertical axis of rotation;

inspecting images in the image sequence for melt surface feature points, as projected into the images, at a number, n, of feature detection locations, the feature detection locations configured with respect to each other at image positions based on an expected shift in image location of a given melt surface feature point between two consecutive images in the image sequence;

mapping detected melt surface feature points from a perspective-view image projection to an orthogonal-view projection of the melt surface; and

identifying as valid melt surface feature points, comprising contour points of solid-phase crystal regions on the surface of the melt, those feature points detected in a specified minimum number of the n feature detection locations each in a different image of the image sequence.

4. The method of claim 3 further comprising analyzing the solid-phase crystal region contour points to determine solid-phase crystal region area.

5. The method of claim 3 further comprising analyzing the solid-phase crystal region contour points to determine an extent of the melt surface that is free of the solid-phase crystal.

6. The method of claim 4 wherein inspecting images for melt surface feature points comprises inspecting images for a solid-phase crystal region nucleating at the ingot.

7. The method of claim 4 wherein inspecting images for melt surface feature points comprises inspecting images for a solid-phase crystal region nucleating at a wall of a crucible in which the melt is contained.

8. The method of claim 7 wherein inspecting images for melt surface feature points further comprises inspecting images for a solid-phase crystal region nucleating at the ingot.

9. The method of claim 8 further comprising analyzing the solid-phase crystal region contour points to determine an extent of the melt surface between solid-phase crystal regions nucleating at the ingot and nucleating at the crucible that is free of solid-phase crystal.