Research On Polycrystalline Silicon Wafer Quality Inspection System Based On Machine Vision

First,the overall framework of crystalline silicon defect detection system is designed: the characteristics of crystalline silicon production are analyzed,and the hardware selection of visual acquisition system,such as the light source,camera,lens etc.,are established.The overall framework of crystalline silicon defect detection is established.Then,a crystalline silicon defect detection algorithm based on machine vision is developed: The dataset containing non-defective and defective crystalline silicon samples is established,and the defect characteristics of crystalline silicon in the production process are analyzed.The sample images are preprocessed by Image segmentation based on Markov Random Field(MRF)theory and bilateral filtering algorithm.The feature vectors of defective crystalline silicon are extracted from preprocessed images.An improved support vector machine is developed to classify the crystalline silicon dataset.Finally,the experimental platform of defective crystalline silicon detection system is built.The defect crystalline silicon detection algorithm based on improved support vector machine(SVM)is verified.The experimental results show that the improved SVM algorithm based on Sequential Minimal Optimization(SMO)can efficiently detect and classify the dataset of non-defective and defective crystalline silicon,and the detection accuracy is over 98%,which can meet the requirements of real-time defect detection in the crystalline silicon production line.There are 43 figures,10 forms and 83 references in this thesis.