Method and system for detecting defects on a printed circuit board

• US 6,771,807 B2
• Filed: 01/18/2001
• Issued: 08/03/2004
• Est. Priority Date: 01/18/2000
• Status: Expired due to Fees

• Current Assignee: SolVision, Inc. (Ametek Incorporated)
• Original Assignee: SolVision, Inc. (Ametek Incorporated)
• Inventors: Bérard, Louis, Gauthier, Jonathan, Cantin, Michel, Coulombe, Alain
• Primary Examiner(s): Bali, Vikkram
• Application Number:
  US09/764,745
• Publication Number:
  US 20010028732A1
• Time in Patent Office: 1,293 Days
• Field of Search: 382/141, 382/143-152, 348/86, 348/87, 348/125, 348/126, 700/95, 700/96, 700/97, 700/117, 700/121
• 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 the PCB image;

providing a computer model for each component on the at least one layer of the PCB, the computer model of each component on the at least one layer including segments;

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

for each detected anomaly, determining if the detected anomaly corresponds to a surface defect;

wherein said edges identifying step includes;

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

thresholding the gradient image to yield a thresholded image;

skeletting the thresholded image; and

finding edges on the skeletted image;

thereby yielding at least one chain of pixels, and wherein said edges identifying step further includes;

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

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

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

whereby, the passage by zero corresponds to edges.

2. The method as recited in claim 1, further comprising correcting a luminosity of the PCB image to yield a corrected image.

3. The method as recited in claim 2, further comprising filtering the corrected image.

4. 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 the PCB image;

providing a computer model for each component on the at least one layer of the PCB, the computer model of each component on the at least one layer including segments, each of the at least one layer of the computer model being aligned with the PCB image and aligned layers being superimposed;

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

for each detected anomaly, determining if the detected anomaly corresponds to a surface defect,

wherein the computer model is a vectorial model including a polygon representation of every component on each of the at least one layer;

the polygon including at least one segment;

the at least one layer of the computer model being superimposed by superimposing each polygon from a first layer with each polygon from a second layer.

5. The method as recited in claim 4, wherein said superimposition is done using the following iterative formula for each layer:

6. The method as recited in claim 4, wherein said anomalies detecting step includes determining, for each layer k of the computer model, a list of polygons T_k, located in narrow regions of the layer k, using the following iterative formula:

7. The method as recited in claim 6, wherein the width of the inter-space is determined by erosion of each of the polygons.

8. The method as recited in claim 4, wherein said anomalies detecting step includes determining, for each of the 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:

9. 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 the PCB image;

providing a computer model for each component on the at least one layer of the PCB, the computer model including contours of each component, and for each contour, a model of a rectangle enclosing the contour;

detecting anomalies on the PCB image by comparing identified edges to the computer model, wherein;

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; and

for each detected anomaly, determining if the detected anomaly corresponds to a surface defect.

10. 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 the PCB image;

providing a computer model for each component on the at least one layer of the PCB, the computer model of each component on the at least one layer including segments;

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

for each detected anomaly, determining if the detected anomaly corresponds to a surface defect;

wherein said anomalies detecting step includes using the computer model to create an image of the computer model and comparing the PCB image with the image of the computer model and further includes detecting edges in the PCB image that are in excess relative to the segments in the computer model;

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

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

11. The method as recited in claim 10, wherein the distance between the first segment and the opposite segment is computed as an effective spacing therebetween when more than one anomaly is located between the first segment and the opposite segment.

12. The method as recited in claim 11, wherein the anomalies include overlapping and non-overlapping anomalies;

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

13. The method as recited in claim 10, wherein the predetermined distance is defined by a rectangle that encloses the detected anomaly.

14. The method as recited in claim 13, wherein the enclosing rectangle is dilated.

15. The method as recited in claim 13, further including verifying if the enclosing rectangle encloses another anomaly and, if so, redefining a rectangle that encloses the another anomaly.

16. The method as recited in claim 15, wherein said rectangle redefining step is iterative.

17. The method as recited in claim 10, wherein the threshold is in a form of a ratio.