Defect Recognition Algorithm Based on Curvelet Moment and Support Vector Machine

摘要

In this paper, a new recognition algorithm based on curvelet moment and support vector machine(SVM) is proposed for chip defect recognition. The proposed recognition method is implemented through a reference comparison method. First the defect regions of chips are extracted through preprocessing, and then the curvelet moment feature of the defect region is computed as the input of SVM classifier, the output of the trained SVM classifier is the result of defect recognition. The algorithm combines the good properties of curvelet moment and SVM classifier, the former can provide multi-scale, local details and orientation information of the defect region, and the latter is suitable to solve the small samples, nonlinear and high dimensions pattern recognition problem. Experimental results show that the algorithm has higher recognition rate compared with PCA based method and can solve the complex defects recognition problem effectively.