PCB Fault Detection Using Image Processing

摘要

The importance of the Printed Circuit Board inspection process has been magnified by requirements of the modern manufacturing environment where delivery of 100% defect free PCBs is the expectation. To meet such expectations, identifying various defects and their types becomes the first step. In this PCB inspection system the inspection algorithm mainly focuses on the defect detection using the natural images. Many practical issues like tilt of the images, bad light conditions, height at which images are taken etc. are to be considered to ensure good quality of the image which can then be used for defect detection. Printed circuit board (PCB) fabrication is a multidisciplinary process, and etching is the most critical part in the PCB manufacturing process. The main objective of Etching process is to remove the exposed unwanted copper other than the required circuit pattern. In order to minimize scrap caused by the wrongly etched PCB panel, inspection has to be done in ear71y stage. However, all of the inspections are done after the etching process where any defective PCB found is no longer useful and is simply thrown away. Since etching process costs 0% of the entire PCB fabrication, it is uneconomical to simply discard the defective PCBs . In this paper a method to identify the defects in natural PCB images and associated practical issues are addressed using Software tools and some of the major types of single layer PCB defects are Pattern Cut, Pin hole, Pattern Short, Nick etc., Therefore the defects should be identified before the etching process so that the PCB would be reprocessed. In the present approach expected to improve the efficiency of the system in detecting the defects even in low quality images