Design And Research Of The Defect Detection System For FPC Based On AOI

In recent years, FPC (Flexible Printed Circuit) have been widely used in the aerospace,military, mobile communications or other fields. But most manufacturers in our country aremainly using manual inspection or semi-automated detection technology to detect FPC，inwhich need a large amount manpower devotion but in low detection efficiency. AOI(Automatic Optic Inspection, AOI) technology with the advantages of high speed,non-contact, high accuracy, which can meet the needs of FPC detection and has animportant engineering significance.This paper has been supported by Important National Science&Technology SpecificProjects,"863"Program and Guangzhou Nansha district economic and information bureauof science and technology.The system uses white LED and array CCD camera to acquirethe board image by scanning FPC automatically. The image acquired will be send to thedefect detection module after a series of preprocessing. Then according to the standard FPCimage information, a new kind of algorithm based on the regularization parameters will beused to detect the defects of the image. At last, the detection result will be collected basedon the allowable threshold. The research focused on the system are mainly concentrated inthe following three aspects:i. On the basis of analyzing the research both at home and abroad, while combiningthe actual application requirements of project equipment, this article discusses thetechnology selection of program systems from both hardware and software;ii. Through in-depth study of the existing circuit board detection algorithms,according to the unique characteristics of FPC boards, a new defect detection method hasbeen put forward. In this way, the image information will be converted into a set ofregularization parameters after the image preprocessing. This method can detect thecommon defects of FPC boards, such as scratches, traces too wide, open circuit and so on,meet the needs of high-speed, high-reliability requirements of automatic detection, whichcan make up for the deficiencies of traditional algorithms; iii. According to the improved algorithm above-mentioned, the FPC defect detectionsystem has been preliminary realized under the windows operating system, including thestandard image generation and image detection, parameter setting, etc. By means ofcarrying the detection test onto10FPC boards which has different types and numbers ofdefects, the stability and reliability of the system is verified. Experimental results show thatcompared with the traditional way of artificial detection, this system can realize thedetection with the line width of15μm, the identification accuracy of3μm and therecognition rate of0.5s/FOV. A full inspection process for the size of504mmx248mm FPCboard only needs10s, that speed is six times higher more than manual visual inspection,having the very high application value.