Study On Automatic Optical Inspection System For SMT Solder Joints Based On Machine Vision

With the development of electronic products directing to high density, thin separation and low defect ration, its inspection requirement is higher on aspects of precision, efficiency, universal and intelligence etc. Image inspection technology based on machine vision is a novel inspection method, which combines optic, photoelectricity, digital image processing, information processing with computer vision. The technology can meet the demand of electronic products inspection.The paper applies image inspection technology to surface mounting technology (SMT) solder joints inspection. Machine vision and SMT products inspection methods are described mainly. The solution of automatic optical inspection (AOI) to SMT production quality inspection is introduced. An automatic optical inspection system for the diagnosis of solder joints defects on printed circuits boards assembled in surface mounting technology is presented. The diagnosis is handled as a pattern recognition problem with a back-propagation (BP) neural network approach. The solder joints image are captured by using only one layer of tiered light. Four features are extracted from the solder joints image, which are segmented four regions using three normalized partition curves. Three types of solder joints have been classified according to the amount of solder: insufficient, acceptable and excessive solders. The experimental results show that the classification correctness reaches 99.2% and the method possesses high practical value.Study the robustness against changes in threshold values and variation in partition curves of four region shows that the present approach is very robust. The paper takes other classifying methods, such as pattern similarity measure,bayes classifier and discriminate function classifier etc. The results of experiment show that the BP neural network classifier is the most greatest.