Method for fast, robust, multi-dimensional pattern recognition

• US 8,363,942 B1
• Filed: 12/31/2004
• Issued: 01/29/2013
• Est. Priority Date: 07/13/1998
• Status: Expired due to Fees

• Current Assignee: Cognex Technology & Investment Corporation (Cognex Corporation)
• Original Assignee: Cognex Technology & Investment Corporation (Cognex Corporation)
• Inventors: Silver, William M., McGarry, E. John, Nichani, Sanjay, Wagman, Adam
• Primary Examiner(s): Koziol, Stephen R
• Application Number:
  US11/027,962
• Time in Patent Office: 2,951 Days
• Field of Search: 382/118, 382/141, 382/181, 382/190, 382/197, 382/209, 382/279, 382/266, 348/92, 348/125
• US Class Current: 382/181
• CPC Class Codes: G06F 18/22 Matching criteria, e.g. proximity measures
• CPC Class Codes: G06V 10/469 Contour-based spatial representations, e.g. vector-coding
• CPC Class Codes: G06V 10/752 Contour matching

Abstract

Disclosed is a method for determining the absence or presence of one or more instances of a predetermined pattern in an image, and for determining the location of each found instance within a multidimensional space. A model represents the pattern to be found, the model including a plurality of probes. Each probe represents a relative position at which a test is performed in an image at a given pose, each such test contributing evidence that the pattern exists at the pose. The method further includes a comparison of the model with a run-time image at each of a plurality of poses. A match score is computed at each pose to provide a match score surface. Then, the match score is compared with an accept threshold, and used to provide the location any instances of the pattern in the image.

Claims

1. In probe-based pattern matching, a method for locating patterns, the method comprising:

providing a database that comprises;

a probe-based pattern model having;

a plurality of positive probes at selected points along a boundary of the pattern,

at least one imaginary straight segment parallel to a segment of the boundary of the pattern, and

a plurality of negative probes at selected points along the imaginary straight segment, each negative probe having a negative weight; and

providing a computer programmed to perform the step of;

using the probe-based pattern model, and probe-based pattern matching, to locate at least one instance of the pattern in an image.

2. The method of claim 1, wherein the plurality of positive probes at selected points along a boundary of the pattern were placed according to a method including:

using the computer to run a program to perform the steps of;

determining an arc position for each boundary point along a segment of the boundary, starting with zero at a first boundary point at a first end of the segment, and increasing while moving away from the first end by an amount equal to the distance between the boundary points along the segment;

determining the total arc length of the segment as being the arc position of a boundary point most distal from the first boundary point of the segment;

determining a probe spacing value along the segment; and

determining a target number of probes to be placed along the segment using the total arc length of the segment, and the probe spacing value.

3. The method of claim 2, wherein determining a probe spacing value along the segment includes:

selecting a probe spacing value along the segment such that the probe spacing value is no less than 0.5 pixels, and such that the probe spacing value is no more than 4.0 pixels.

4. The method of claim 2, wherein determining a target number of probes to be placed along the segment includes:

computing a ratio of the arc length of the segment to the probe spacing value along the segment;

computing a floor function of the ratio; and

adding 1.

5. The method of claim 2, after determining a target number of probes to be placed along the segment, further comprising:

using fewer than the target number of probes to ensure that the spacing does not fall below a predetermined lower limit.

6. The method of claim 2, after determining a target number of probes to be placed along the segment, further comprising:

using more than the target number of probes to ensure that the spacing does not exceed a predetermined upper limit.

7. The method of claim 6, wherein a position of each probe is interpolated from positions of surrounding boundary points.

8. The method of claim 6, wherein a direction of each probe is interpolated from directions of surrounding boundary points.

9. The method of claim 1, wherein the plurality of positive probes at selected points along a boundary of the pattern were placed according to a method including:

using the computer to run a program to perform the steps of;

identifying at least one long segments of the boundary of the pattern;

identifying at least one short segments of the boundary of the pattern;

associating a long positive weight with each positive probe along each long segment;

associating a short positive weight with each positive probe along each short segment; and

setting the value of each short positive weight as being greater than the value of each long positive weight.

10. The method of claim 1, wherein the plurality of positive probes at selected points along a boundary of the pattern were placed according to a method including:

using the computer to run a program to perform the steps of;

identifying at least one long segments of the boundary of the pattern;

identifying at least one short segments of the boundary of the pattern; and

placing more probes per unit length on each short segment than on each long segment.

11. The method of claim 1, wherein the plurality of negative probes at selected points along the imaginary straight segment, each negative probe having a negative weight, were placed according to a method including:

using the computer to run a program to perform the steps of;

making the magnitude of each negative weight greater than the magnitude of positive weights on the at least one straight segment that was extended to provide the imaginary straight segment.

12. The method of claim 1, wherein the plurality of negative probes at selected points along the imaginary straight segment, each negative probe having a negative weight, were placed according to a method including:

using the computer to run a program to perform the steps of;

making the number of negative probes along the imaginary straight segment exceed the number positive probes on the at least one straight segment that was extended to provide the imaginary straight segment.

13. The method of claim 1, wherein the plurality of positive probes at selected points along a boundary of the pattern were placed according to a method including:

using the computer to run a program to perform the steps of;

identifying a pair of long segments;

identifying a pair of short segments;

associating a long positive weight with each positive probe along each long segment;

associating a short positive weight with each positive probe along each short segment; and

setting the value of each short positive weight as being greater than the value of each long positive weight.

14. The method of claim 1, wherein the plurality of positive probes at selected points along a boundary of the pattern were placed according to a method including:

using the computer to run a program to perform the steps of;

identifying a pair of long segments;

identifying a pair of short segments; and

placing as many probes on each long segment as there are probes placed on each corresponding short segment.

15. The method of claim 1, wherein the plurality of negative probes at selected points along the imaginary straight segment, each negative probe having a negative weight, were placed according to a method including:

using the computer to run a program to perform the steps of;

making the magnitude of each negative weight greater than the magnitude of each positive weights on a straight segment.

16. The method of claim 1, wherein the plurality of negative probes at selected points along the imaginary straight segment, each negative probe having a negative weight, were placed according to a method including:

using the computer to run a program to perform the steps of;

making the number of negative probes along the imaginary straight segment exceed the number positive probes on a straight segment.

17. The method of claim 13, wherein only two pairs of long segments are identified.

18. The method of claim 13, wherein only two pairs of short segments are identified.

19. The method of claim 1, wherein the probe-based pattern model has 36 probes.

20. The method of claim 1, wherein weights of probes of the short segments are twice the value of weights of probes of the long segments.

21. The method of claim 10 wherein the step of placing more probes per unit length on each short segment than on each long segment include placing as many probes on each long segment as there are probes placed on each short segment.

22. In probe-based pattern matching, an apparatus for locating patterns, the apparatus comprising:

a database including a probe-based pattern model, the model including;

(i) a plurality of positive probes at selected points along a boundary of the pattern;

(ii) at least one imaginary straight segment parallel to a segment of the boundary of the pattern; and

(iii) a plurality of negative probes at selected points along the imaginary straight segment, each negative probe having a negative weight; and

a computer readable memory including a program to perform the steps of;

using the probe-based pattern model, and probe-based pattern matching, to locate at least one instance of the pattern in an image.

23. The apparatus of claim 22 wherein the pattern includes at least one long segment of the boundary and at least one short segment of the boundary of the pattern, a long positive weight associated with each positive probe along each long segment and a short positive weight associated with each positive probe along each short segment, where the value of each short positive weight is being greater than the value of each long positive weight.

24. The apparatus of claim 22 wherein the pattern includes at least one long segment of the boundary and at least one short segment of the boundary of the pattern and the probes include more probes per unit length on each short segment than on each long segment.

25. The apparatus of claim 22, wherein a mimicked-shape, having part of the boundary and the at least one imaginary straight segment, match a feature in the image that is not in an instance of the pattern in the image.

26. The apparatus claim 22, wherein the program in computer readable memory places a plurality of positive probes at selected points along a boundary of the pattern, the program further comprising:

computing an arc position for each boundary point along a segment of the boundary, starting with zero at a first boundary point at a first end of the segment, and increasing while moving away from the first end by an amount equal to the distance between the boundary points along the segment;

computing the total arc length of the segment as being the arc position of a boundary point most distal from the first boundary point of the segment;

computing a probe spacing value along the segment; and

calculating a target number of probes to be placed along the segment using the total arc length of the segment, and the probe spacing value.

27. The apparatus claim 22, wherein the magnitude of each negative weight being greater than the magnitude of one of the positive weights on the at least one straight segment that was extended to provide the imaginary straight segment.