Font Size: a A A

Electronic Assembly Defect Inspection Base On Digital Image Processing Technology

Posted on:2011-11-05Degree:MasterType:Thesis
Country:ChinaCandidate:L YangFull Text:PDF
GTID:2218330338450154Subject:Electronics and Communications Engineering
Abstract/Summary:PDF Full Text Request
With the development of electronics industry, surface mount technology is becoming the major technology in electronic assembly process. Because the quality of the assembly will impact both mechanical and electrical performance, the defect inspection is becoming more and more important.Firstly, the features of surface mount technology and some assembly defects are introduced. And some normal inspection methods and the automatic optical inspection (AOI) which is base on digital image processing technology are amplified. And then, the development of automatic optical inspection researching is summarized.Secondly, after the analysis of PCBA images and the major noise disturbance types, the methods of digital image pretreatment are studied, which include image sharpen, contrast improvement and noise disturbance restraining. At the same time, a new method for image locating and adjusting is given, which is base on HSI color mode. Through extracting the mark point on the PCBA, a system of coordinates is established and the image locating and adjusting are realized.Thirdly, two defect inspections are introduced. One is wrong component inspection, and the other is RCL component solder defect inspection. About wrong component inspection, component silkscreen features, character image pretreatment and features extracting are amplified, and then a BP network is used to realize character recognition. About RCL component solder defect inspection, PCBA image features and image features selecting are studied. According to the actual production, a new method named two stages inspection is given. At the first stage, it uses a single image feature to input into a perceptron network and checks out many good samples with low losing rate. Then at the second stage, it extracts the principle component from multi image features and inputs them into a BP network, and then identifies different defect mode. Comparing with the normal defect inspection methods, two stages inspection method performs wonderfully in inspection yield and efficiency.Lastly, a summarization and the view of next step are given.
Keywords/Search Tags:Automatic Optical Inspection, Digital Image Processing, Neural Network, Principle Component Analysis, Defect Inspection
PDF Full Text Request
Related items