Method and system for detecting defects on a printed circuit board

• US 20010028732A1
• Filed: 01/18/2001
• Published: 10/11/2001
• Est. Priority Date: 01/18/2000
• Status: Active Grant

• Current Assignee: SolVision, Inc. (Ametek Incorporated)
• Original Assignee: SolVision, Inc. (Ametek Incorporated)
• Inventors: Cantin, Michel, Coulombe, Alain, Berard, Louis, Gauthier, Jonathan
• Granted Patent: US 6,771,807 B2
• Time in Patent Office: Days
• US Class Current: 382/149
• CPC Class Codes: G06T 2207/30141 Printed circuit board [PCB]
• CPC Class Codes: G06T 7/0006 using a design-rule based approach

Abstract

A method and a system for detecting surface defects on electronic circuits, such as Printed Circuit Boards (PCB), are described herein. The method first comprises identifying contours on a digital image of the PCB. Then anomalies are detected on the PCB image by comparing the identified contours to contours on a vectorial model of the PCB. Each detected anomaly is compared to manufacturing data to verify if it corresponds to a defect.

Claims

1. A method for detecting surface defects on a Printed Circuit Board (PCB) having at least one layer including components, said method comprising:

providing a digital image of the PCB;

identifying edges on said PCB image;

providing a corresponding computer model for each component on said at least one layer of the PCB;

detecting anomalies on the PCB image by comparing said identified edges to said computer model; and

for each of said detected anomalies, determining if said detected anomaly corresponds to a surface defect.

2. A method for detecting surface defects on a PCB as recited in

claim 1

, wherein said computer model of each component on said at least one layer including segments.

3. A method for detecting surface defects on a PCB as recited in

claim 1

, wherein said edges identifying step includes:

computing the numerical gradient on said PCB image to yield a gradient image;

thresholding said gradient image to yield a thresholded image;

skeletting said thresholded image; and

finding edges on the skeletted image.

4. A method for detecting surface defects on a PCB as recited in

claim 3

, wherein the edges finding step yields at least one chain of pixels and includes:

saving each of said at least one of said chain of pixels of the skeletted image as a vector of pixel co-ordinates;

using said pixel co-ordinates to estimate the second derivative of said at least one chain of pixels; and

estimating the co-ordinates of the passage by zero of said second derivatives;

whereby, the passage by zero correspond to said edges.

5. A method for detecting surface defects on a PCB as recited in

claim 3

, further comprising correcting the luminosity of the PCB image.

6. A method for detecting surface defects on a PCB as recited in

claim 5

, further comprising filtering the corrected image;

7. A method for detecting surface defects on a PCB as recited in

claim 1

, wherein said provided computer model includes a vectorial model for each of said layer.

8. A method for detecting surface defects on a PCB as recited in

claim 7

, wherein said vectorial model includes a polygon representation of every components on each layer.

9. A method for detecting surface defects on a PCB as recited in

claim 2

, wherein each of said at least one layer of said computer model being aligned with said PCB image before the anomalies detecting step.

10. A method for detecting surface defects on a PCB as recited in

claim 9

, wherein at least one reference point is selected on the PCB image for each of said at least one layer of the computer model;

said reference point is used to align said corresponding layer with said PCB image.

11. A method for detecting surface defects on a PCB as recited in

claim 9

, wherein transformations are applied on every segment of each layer of said computer model to align said each of layer with said PCB image so as to compensate for at least one of rotation, translation, scaling and bad perspective of said each layer.

12. A method for detecting surface defects on a PCB as recited in

claim 9

, wherein said aligned layers are superimposed.

13. A method for detecting surface defects on a PCB as recited in

claim 12

, wherein said provided computer model is a vectorial model including a polygon representation of every components on each layer;

said polygon including at least one segment;

said layers of said computer model are superimposed by superimposing each polygon from each layer with each polygon from the other layer.

14. A method for detecting surface defects on a PCB as recited in

claim 13

, wherein said superimposition is done using the following iterative formula for each layer:

Pv_k+1=(Pv_k−

P_k+1)∪

P_K+1

where

P_k is a list of vectorial representations of segments in the current layer;

Pv_k is a list of vectorial representations of segments produced by the superimposition of the previous layers;

Pv_k+1 is a list of vectorial representations of segments produced by the superimposition a layer with the superimposition of the previous layers; and

Pv₁₌P₁

15. A method for detecting surface defects on a PCB as recited in

claim 1

, wherein said computer model includes contours of each component.

16. A method for detecting surface defects on a PCB as recited in

claim 15

, wherein said computer model further includes, for each contour, a model of a rectangle enclosing said contour.

17. A method for detecting surface defects on a PCB as recited in

claim 16

, wherein said anomalies detecting step includes:

a) for each pixel of the PCB image, verifying if a contour among said contours of each component has its corresponding rectangle enclosing said pixel;

if not, considering said pixel part of an anomaly;

if so, b) verifying if said pixel is part of said contour having its corresponding rectangle enclosing said pixel;

if not, considering said pixel part of an anomaly.

18. A method for detecting surface defects on a PCB as recited in

claim 1

, wherein said anomalies detecting step includes using said computer model to create an image of said computer model and comparing said PCB image with said image of said computer model.

19. A method for detecting surface defects on a PCB as recited in

claim 2

, wherein the determination of defects includes comparing said detected anomalies to design specifications.

20. A method for detecting surface defects on a PCB as recited in

claim 19

, wherein the step of detecting anomalies on the PCB image further including detecting edges in the PCB image that are in excess relative to said segments in said computer model;

for each of said detected anomalies, determining defects by a) determining a first segment in said computer model that is positioned within a predetermined distance of one of said detected anomalies;

b) for each segment opposed to said opposed segment, determining if the distance between said first segment and said segment opposed to said first segment is within a predetermined threshold.

21. A method for detecting defects on a PCB as recited in

claim 20

, wherein said distance between said first segment and said segment opposed to said first segment is computed as the effective spacing therebetween when more than one anomalies is located between said first segment and said opposed segment.

22. A method for detecting defects on a PCB as recited in

claim 21

, wherein said anomalies include overlapping and non-overlapping anomalies;

said effective spacing is computed as the sum of intervals between all non-overlapping anomalies and regrouped overlapping anomalies.

23. A method for detecting defects on a PCB as recited in

claim 20

, wherein said predetermined distance is defined by a rectangle that encloses said detected anomaly.

24. A method for detecting defects on a PCB as recited in

claim 23

, wherein said enclosing rectangles being dilated.

25. A method for detecting defects on a PCB as recited in

claim 23

, further including verifying if said enclosing rectangle encloses another anomaly and, if so, redefining a rectangle that encloses said another anomaly.

26. A method for detecting defects on a PCB as recited in

claim 25

, wherein said rectangle redefining step is iterative.

27. A method for detecting defects on a PCB as recited in

claim 20

, wherein said threshold is in the form of a ratio.

28. A method for detecting surface defects on a PCB as recited in

claim 13

, wherein said anomalies detecting step includes determining, for each layer k of said computer model, a list of polygons T_k, located in narrow regions of said layer k, using the following iterative formula:

T_k+1=(T_k−

P_k+1)∪

((D(P_k+1)∩

D(Pv_k))−

D(P_k+1∩

Pv_k))

where

P_k+1 is a list of vectorial representations of polygons in the previous layer;

Pv_k is a list of vectorial representations of polygons produced by the superimposition of the previous layers;

Pv_k+1 is a list of vectorial representations of polygons produced by the superimposition a layer with the superimposition of the current layer;

D is a function that causes the dilatation of all polygons in a list of polygons; and

T₁=D(P₁)∩

D(P₁);

two opposite segments in said list of polygons yielding an inter-space having a width;

each of said inter-space corresponding to a defect is said corresponding width is not within a predetermined threshold.

29. A method for detecting surface defects on a PCB as recited in

claim 28

, wherein said width of said inter-space is determined by erosion of each of said polygons.

30. A method for detecting surface defects on a PCB as recited in

claim 13

, wherein said anomalies detecting step includes determining, for each of said at least one layer k, a list of polygons O_k in overlapping regions on the PCB'"'"'s computer model using the following iterative formula:

O_k+1=(O_k−

P_k+1)∪

(P_k+1∩

Pv_k)

where;

P_k+1 is a list of vectorial representations of polygons in the previous layer;

Pv_k is a list of vectorial representations of polygons produced by the superimposition of the previous layers; and

O₁ is nul;

two opposite segments in said list of polygons yielding an inter-space having a width;

each of said inter-space corresponding to a defect is said corresponding width is not within a predetermined threshold.

31. A method for detecting surface defects on a PCB as recited in

claim 30

, wherein said list of polygons includes visible and hidden contours;

said hidden contours being extrapolated using said visible contours.

32. A method for detecting surface defects on a PCB as recited in

claim 1

, further comprising indexing said detected anomalies in a computer file.

33. A method for detecting surface defects on a PCB as recited in

claim 1

, wherein said image is in the form of a bitmap.

34. A method for detecting surface defects on a PCB as recited in

claim 33

, wherein said bitmap includes grey levels.

35. A system for detecting surface defects on a PCB;

said system comprising;

a computer including a model of the PCB and being configured to identify edges on a PCB images, detecting anomalies on the PCB image by comparing said identified edges to said computer model, and for each detected anomaly, determining if said detected anomaly corresponds to a surface defect;

an illumination assembly connected to said computer to provide illumination on the PCB;

a frame grabber connected to said computer;

a camera connected to said frame grabber to take an image of the PCB; and

a positioning system connected to said frame grabber.

36. A system as recited in

claim 35

, wherein said computer is further configured so as to control the illumination assembly.

37. A system as recited in

claim 35

, wherein said positioning system is connected to said frame grabbed via an axis controller and a servo controller.

38. A system as recited in

claim 35

, further comprising an image processing card for interfacing said frame grabber and said computer.

39. A system as recited in

claim 35

, wherein said camera is a Charge-Coupled Device (CCD).

40. A system as recited in

claim 39

, wherein said CCD is configured to operate in linear mode.

41. A system as recited in

claim 39

, wherein said CCD is configured to operate in matrix mode.

42. A system as recited in

claim 35

, wherein said positioning system includes at least one servomotor.