Methods and apparatuses for generating a model of an object from an image of the object

• US 6,813,377 B1
• Filed: 07/31/2000
• Issued: 11/02/2004
• Est. Priority Date: 08/06/1999
• Status: Expired due to Fees

• Current Assignee: Cognex Corporation
• Original Assignee: Cognex Corporation
• Inventors: Gopalakrishnan, Venkat, Montillo, Albert, Bachelder, Ivan
• Primary Examiner(s): Mehta, Bhavesh M.
• Assistant Examiner(s): CHAWAN, SHEELA C
• Application Number:
  US09/630,004
• Time in Patent Office: 1,555 Days
• Field of Search: 382/146, 382/145, 382/149, 382/285, 382/147, 382/144, 382/150, 382/106, 382/141, 382/151, 382/191, 382/207, 382/274, 250/208.1, 250/235, 250/559.22, 250/559.34, 250/559.36, 250/559.37, 250/306, 356/394, 356/404, 356/398, 356/237.5, 356/418, 348/87, 348/126
• US Class Current: 382/146
• CPC Class Codes: G06F 18/28 Determining representative reference patterns, e.g. by averaging or distorting; Generating dictionaries
• CPC Class Codes: G06V 30/1478 of characters or characters lines

Abstract

A method and an apparatus are disclosed for generating a model of an object from an image of the object. First, an orientation of the object in the image is determined through the generation of, and subsequent evaluation of, at least a portion of a frequency response of the image. Thereafter, the orientation is used to gauge the object. The gauging provides the necessary dimensional information that becomes a part of the geometrical model of the object. An embodiment is disclosed that generates a geometrical model of a leaded object from the image of the leaded object.

Claims

1. A method for generating a geometrical model of an object from an intensity image of the object, the method comprising:

generating at least a portion of a two-dimensional frequency response of the intensity image, the at least a portion of a two-dimensional frequency response being a frequency response to more than one frequency;

representing the at least a portion of a two-dimensional frequency response within a frequency space, the at least a portion of a frequency response providing features arranged in a spatial-frequency pattern within the frequency space;

finding an orientation of the spatial-frequency pattern to provide the orientation of the object;

gauging the object using the orientation so as to produce geometric-dimensional information; and

constructing the geometrical model of the object using the geometric-dimensional information.

2. The method of claim 1, further comprising:

using the geometrical model as an initial input into a more precise modeling procedure.

3. The method of claim 1, wherein the object has object features, the object features being within a region of the intensity image, and wherein gauging further includes:

gauging the object features within the region using the orientation so as to produce the geometric-dimensional information for the object features;

the method further comprising;

segmenting the region before gauging.

4. The method of claim 3, wherein segmenting the region further includes:

segmenting the region of the object using the orientation.

5. The method of claim 3, wherein segmenting the region further includes:

dividing at least a portion of the intensity image into sub-regions, each of the sub-regions having an origin;

generating a frequency characteristic along the orientation for each of the sub-regions; and

examining the frequency characteristic of each of the sub-regions to identify similar sub-regions, the similar sub-regions taken together providing the region.

6. The method of claim 3, wherein segmenting includes:

segmenting a plurality of regions;

the method further comprising;

ignoring at least one region of the plurality of regions having an aspect in conflict with a validity norm.

7. The method of claim 6, wherein the validity norm is a predetermined area threshold, and wherein ignoring further includes:

ignoring at least one region of the plurality of regions having an area above the predetermined area threshold.

8. The method of claim 1, wherein the object is a leaded object having lead set regions, and having leads within the lead set regions, and wherein the method further comprises:

segmenting the lead set regions before gauging.

9. The method of claim 8, wherein segmenting the lead set regions further includes:

thresholding the intensity image at a plurality of multiple thresholds so as to produce a group of the lead set regions at each threshold of the plurality of multiple thresholds; and

segmenting the group of the lead set regions containing the lead set regions that are substantially rectangular.

10. The method of claim 8, further comprising:

ignoring at least one lead set region of the lead set regions having an aspect in conflict with a validity norm.

11. The method of claim 10, wherein the validity norm is a substantially rectangular shape, and wherein ignoring further includes:

ignoring at least one lead set region of the lead set regions with a shape unlike the substantially rectangular shape.

12. The method of claim 8, wherein gauging the object further includes:

applying an edge detection tool using the orientation so as to produce a lead width, and a lead pitch for each of the leads within at least one lead set region of the lead set regions.

13. The method of claim 8, wherein gauging the object further includes:

applying an edge detection tool using the orientation to produce a lead length and a foot length for the leads within at least one lead set region of the lead set regions.

14. The method of claim 1, wherein generating further includes:

generating the at least a portion of a two-dimensional frequency response of the intensity image by applying a frequency analysis tool to the intensity image.

15. The method of claim 14, wherein generating the at least a portion of a two-dimensional frequency response of the intensity image includes:

generating at least a portion of a magnitude of a two-dimensional discrete Fourier transform of the intensity image to provide the at least a portion of a two-dimensional frequency response.

16. The method of claim 14, wherein generating the at least a portion of a two-dimensional frequency response of the intensity image includes:

generating at least a portion of a two-dimensional discrete cosine-transform of the intensity image to provide the at least a portion of a two-dimensional frequency response.

17. The method of claim 1, wherein representing further includes:

representing the at least a portion of a two-dimensional frequency response as a logarithmically scaled frequency response within the frequency space.

18. The method of claim 1, wherein the frequency space is a frequency image, and wherein representing the at least a portion of a two-dimensional frequency response further includes:

scaling the at least a portion of a two-dimensional frequency response using a scaling function so as to enhance high frequency responses within the at least a portion of the two-dimensional frequency response to provide a scaled frequency response; and

mapping the scaled response by gray scale on the frequency image.

19. The method of claim 1, wherein finding further includes:

applying an angle finding means to the frequency space to provide an angle of the spatial-frequency pattern.

20. The method of claim 1, wherein finding further includes:

identifying the spatial-frequency pattern within the frequency space; and

finding the orientation of the spatial-frequency pattern.

21. The method of claim 1, wherein the features form a plurality of spatial-frequency patterns within the frequency space, and wherein finding further includes:

finding the orientation of the plurality of spatial-frequency patterns.

22. The method of claim 1, wherein the features form a plurality of spatial-frequency patterns within the frequency space, and wherein finding further includes:

identifying a dominant spatial-frequency pattern among the plurality of spatial-frequency patterns; and

finding the orientation of the dominant spatial-frequency pattern.

23. The method of claim 22, wherein identifying further includes:

identifying as the dominant spatial-frequency pattern, one spatial-frequency pattern of the plurality the spatial-frequency patterns that includes a greatest number of the features.

24. The method of claim 1, wherein the orientation of the object in the intensity image is at a constant offset from the orientation of the spatial-frequency pattern in the frequency space.

25. The method of claim 24, wherein the constant offset substantially equals zero.

26. The method of claim 24, wherein the orientation of the object substantially equals an orientation angle of a surface mount object from an axis of the intensity image.

27. The method of claim 1, wherein the spatial-frequency pattern includes a line, the line having, a line angle, and wherein the orientation of the object is an object angle, the object angle having a constant offset from the line angle.

28. The method of claim 27, wherein the orientation of the object is defined by an orientation angle of a feature on the object.

29. The method of claim 1, wherein finding the orientation of the spatial-frequency pattern further includes:

processing the frequency space as a second image to find the orientation of the spatial-frequency pattern therein.

30. The method of claim 1, wherein representing the at least a portion of a two-dimensional frequency response further includes:

interpreting the features of the at least a portion of a two-dimensional frequency response within the frequency space as image features in a second intensity image.

31. The method of claim 1, wherein finding the orientation of the spatial-frequency pattern further includes:

conducting a pattern analysis on the features.

32. The method of claim 1, wherein the features in the frequency space are formed from uv data and finding the orientation further includes:

interpreting the uv data as a second intensity image to find the orientation of the spatial-frequency pattern therein.

33. A method for generating a geometrical model of an object from an image of an object, the method comprising:

acquiring the intensity image of the object;

generating at least a portion of a two-dimensional frequency response of the intensity image by applying a frequency analysis tool to the intensity image, the at least a portion of a two-dimensional freuency response being a frequency response to more than one frequency, frequencies within the at least a portion of a frequency response forming a spatial-frequency pattern;

finding an orientation of the spatial-frequency pattern to provide the orientation of the object;

gauging the object using the orientation so as to produce geometric-dimensional information; and

constructing the geometrical model of the object using the geometric-dimensional information.

34. The method of claim 33, wherein generating further includes:

representing the at least portion of a two-dimensional frequency response within a frequency space, features within the frequency space forming the spatial-frequency pattern.

35. The method of claim 34, wherein the frequency space is a frequency image, and wherein representing the at least a portion of a two-dimensional frequency response further includes:

scaling the at least a portion of a two-dimensional frequency response using, a scaling function so as to enhance high frequency respones within the at least a portion of the two-dimensional frequency response to provide a scaled frequency response; and

mapping the scaled response by gray scale on the frequency image.

36. The method of claim 33, wherein generating the at least a portion of a two-dimensional frequency response of the intensity image includes:

generating at least a portion of a magnitude of a two-dimensional discrete Fourier transform of the intensity image to provide the at least a portion of a two-dimensional frequency response.

37. The method of claim 33, wherein the orientation of the object in the intensity image is at a constant offset from the orientation of the spatial-frequency pattern in the frequency space.

38. The method of claim 33, wherein the spatial-frequency pattern includes a line, the line having a line angle, and wherein the orientation of the object is an object angle, the object angle having a constant offset from the line angle.

39. The method of claim 38, wherein the orientation of the object is defined by an orientation angle of a feature on the object.

40. The method of claim 33, wherein the object has object features, the object features being within a region of the intensity image, and wherein gauging further includes:

gauging object features within the region using the orientation so as to produce the geometric-dimensional information for the object features;

the method further comprising;

segmenting the region before gauging.

41. The method of claim 40, wherein segmenting further includes:

dividing at least a portion of the intensity image into sub-regions, each of the sub-regions having an origin;

generating a frequency characteristic along the orientation for each of the sub-regions; and

examining the frequency characteristic of each of the sub-regions to identify similar sub-regions, the similar sub-regions taken together providing the region.

42. The method of claim 33, wherein generating the at least a portion of a two-dimensional frequency response further includes:

representing the spatial-frequency pattern in a frequency space; and

wherein finding the orientation further includes;

processing the frequency space as a second intensity image to find the orientation of the spatial-frequency pattern therein.

43. An apparatus for generating a geometrical model of an object from an image of the object, the apparatus comprising:

frequency means adapted to generate at least a portion of a two-dimensional frequency response of the intensity image, the at least a portion of a two-dimensional frequency response being a frequency response to more than one frequency;

a frequency space adapted to receive the at least a portion of a two-dimensional frequency response, the at least a portion of a frequency response providing features arranged in a spatial-frequency pattern within the frequency space;

finding means, adapted to find an orientation of the spatial-frequency pattern to provide the orientation of the object;

gauging means, in cooperation with the finding means, adapted to gauge the object using the orientation so as to produce geometric-dimensional information; and

modeling means, in cooperation with the gauging means, adapted to construct the geometrical model of the object using the geometric-dimensional information.

44. The apparatus of claim 43, wherein the frequency means is further adapted to generate the at least a portion of a two-dimensional frequency response of the intensity image by applying a frequency analysis tool to the intensity image.

45. The apparatus of claim 44, wherein the frequency means is further adapted to generate at least a portion of a magnitude of a two-dimensional discrete Fourier transform of the intensity image to provide the at least a portion of a two-dimensional frequency response.

46. The apparatus of claim 43, wherein the frequency space is a frequency image, the apparatus further comprising:

scaling means, adapted to scale the at least a portion of a two-dimensional frequency response on the frequency space using a scaling function so as to enhance high frequency responses within the at least a portion of the two-dimensional frequency response and to provide a scaled frequency response; and

mapping means, in cooperation with the scaling means, adapted to map the scaled frequency response by grey scale on the frequency image.

47. The apparatus of claim 43, wherein the orientation of the object in the intensity image is at a constant offset from the orientation of the spatial-frequency pattern in the frequency space.

48. The apparatus of claim 47, wherein the orientation of the object is defined by an orientation angle of a feature on the object.

49. The apparatus of claim 43, wherein the spatial-frequency pattern includes a line, the line having a line angle, and wherein the orientation of the object is an object angle, the object angle having a constant offset from the line angle.

50. The apparatus of claim 43, wherein the object has object features, the object features being within a region of the intensity image, and wherein the gauging means is further adapted to gauge the object features within the region using the orientation so as to produce the geometric-dimensional information for the object features;

the apparatus further comprising;

segmenting means, in cooperation the gauging means, adapted to segment the region before gauging.

51. The apparatus of claim 50, wherein the segmenting means further includes:

dividing means adapted to divide at least a portion of the intensity image into sub-regions, each of the sub-regions having an origin;

sub-region frequency means, in cooperation with the dividing means, adapted to generate a frequency characteristic along the orientation for each of the sub-regions; and

examining means, in cooperation with the sub-region frequency means, adapted to examine the frequency characteristic of each of the sub-regions to identify similar sub-regions, the similar sub-regions taken together providing the region.

52. The apparatus of claim 43, wherein the finding means is further adapted to processes the frequency space as a second intensity image to find the orientation of the spatial-frequency pattern.

53. The apparatus of claim 43, wherein the features within the frequency space are image features of a second intensity image.

54. The apparatus of claim 43, wherein the finding means is further adapted to conduct a pattern analysis on the frequency space.