Machine vision method using search models to find features in three dimensional images

• US 6,173,070 B1
• Filed: 12/30/1997
• Issued: 01/09/2001
• Est. Priority Date: 12/30/1997
• Status: Expired due to Term

• Current Assignee: Cognex Corporation
• Original Assignee: Cognex Corporation
• Inventors: Michael, David J., Wallack, Aaron S.
• Primary Examiner(s): Couso, Jose L.
• Assistant Examiner(s): Dang, Duy M.
• Application Number:
  US09/000,583
• Time in Patent Office: 1,106 Days
• Field of Search: 382/141, 382/142, 382/144-154, 382/325, 382/285-287, 382/291, 364/582, 364/491, 364/560, 348/87, 348/126
• US Class Current: 382/145
• CPC Class Codes: G06T 2200/04 involving 3D image data
• CPC Class Codes: G06T 2207/30152 Solder
• CPC Class Codes: G06T 7/001 using an image reference approach
• CPC Class Codes: G06T 7/75 involving models
• CPC Class Codes: G06V 20/64 Three-dimensional objects

Abstract

A system and method for analyzing 3D captured image data of scanned surfaces to locate object of interest, such as solder balls. A 3D model is created of an object of interest, the 3D model including weighted “don'"'"'t care” areas where a match is not required. The 3D models which are used include geometric shapes, including a frustum, and models created from actual 3D image data of real objects. The 3D captured image data is processed to locate objects matching the 3D models, including processing by normalized correlation. Once the objects are located, the system selects data points within the 3D captured image data that are a predetermined distance away from the located objects. These data points are analyzed to determine a surface plane which fits the selected data points, thereby locating the surface in the 3D capture image data.

Claims

1. In a system for analyzing 3D image data obtained by a 3D imaging system, a method for locating objects of interest in said 3D image data comprising:

defining a 3D model of at least one object of interest which includes at least one weighted area indicating an area of said 3D model wherein a match to corresponding areas in said 3D image data is not of interest;

comparing said 3D model to said 3D image data using normalized correlation; and

providing an indication if said 3D model substantially matches part of said 3D image data.

2. The method of claim 1 wherein said 3D image data includes a plurality of data points, and corresponding merit score data provided by said 3D imaging system, said merit score data providing an indication of a confidence in validity in each of said corresponding data points.

3. The method of claim 2 wherein said at least one weighted area is determined to correspond to areas of said 3D model, which, when compared to said 3D image data, will correspond to data points in said 3D image data with low corresponding merit scores.

4. The method of claim 1 wherein said step of comparing said 3D model to said 3D image data uses weighted normalized correlation.

5. In a system for analyzing 3D image data obtained by a 3D imaging system, a method for locating objects of interest in said 3D image data comprising:

defining a 3D model of at least one object of interest;

defining at least one weighted area of said 3D model, said weighted area indicating an area of said 3D model wherein a match to a corresponding area in said 3D image data is not of interest;

comparing said 3D model to said 3D image data; and

providing an indication if said 3D model matches part of said 3D image data.

6. The method of claim 5 wherein said step of comparing said 3D model to said 3D image data includes performing normalized correlation.

7. The method of claim 5 wherein said step of comparing said 3D model to said 3D image data includes performing weighted normalized correlation.

8. The method of claim 5 wherein said step of comparing said 3D model to said 3D image data includes performing regular correlation.

9. The method of claim 5 wherein said at least one weighted area of said 3D model indicates an area of said 3D model wherein any match to corresponding areas in said 3D image data is unnecessary.

10. The method of claim 5 wherein a shape of said 3D model includes a frustum (truncated cone).

11. The method of claim 5 wherein said step of defining a 3D model includes defining a 3D model by processing 3D image data of at least one of said objects of interest.

12. The method of claim 5 wherein said step of defining at least one weighted area of said 3D model includes determining areas of said 3D model which, when compared to said 3D image data, are likely to correspond to parts of said 3D image data which contain erroneous data.

13. The method of claim 5 wherein said step of providing an indication if said 3D model matches part of said 3D image data includes providing a location of said matched object of interest in said 3D image data.

14. The method of claim 13 further including the steps of:

selecting data in said 3D image data that is a predetermined distance away from said location of said matched object of interest in said 3D image data; and

processing said selected data to determine a surface plane which includes a pre-determined portion of said selected data.

15. The method of claim 14 wherein said at least one object of interest includes solder balls to be located and analyzed on the surface of an electrical component, said electrical component being scanned by a 3D image acquisition device to produce said 3D image data.

16. The method of claim 15 wherein said step of processing said selected data to determine a surface plane includes determining a surface of said electrical component.

17. The method of claim 15 further including the step of:

analyzing said located solder balls to determine properties of said solder balls.

18. The method of claim 17 wherein said properties of said solder balls to be determined are selected from the group of:

height, volume, shape, center of mass and coplanarity.

19. The method of claim 14 wherein said step of selecting data in said 3D image data that is a predetermined distance away from said location of said matched object of interest in said 3D image data is performed for a plurality of matched objects in said 3D image data.

20. The method of claim 19 wherein said step of processing said selected data to determine a surface plane includes computing a least square fit on said selected data.

21. A computer system for locating and analyzing objects of interest in 3D image data comprising:

a memory storing said 3D image data; and

a processor comparing a 3D model of a predetermined object of interest to said 3D image data stored in said memory;

said 3D model including at least one weighted area, said at least one weighted area indicating an area of said 3D model wherein a match to said 3D image data is not of interest,

wherein said processor provides an indication if said 3D model matches a part of said 3D image data.

22. The system of claim 21, wherein said processor compares said 3D model to said 3D image data using weighted normalized correlation.

23. The system of claim 21, wherein a shape of said 3D model of a predetermined object of interest includes a frustum (truncated cone).

24. The system of claim 21 wherein said 3D model is created by locating a predetermined object of interest in captured 3D image data, and processing said captured 3D image data including said predetermined object of interest to produce said 3D model.

25. The system of claim 21 wherein said at least one weighted area of said 3D model includes areas of said 3D model which, when compared to said 3D image data, are likely to correspond to parts of said image data which contain erroneous data.

26. The system of claim 21 wherein said processor, upon locating at least one predetermined object of interest in said 3D image data, selects data in said 3D image data that is a predetermined distance away from said located at least one predetermined object of interest, and determines a surface plane which includes a predetermined portion of said selected data.

27. The system of claim 26 wherein said surface plane includes a complex quadric surface.