SIFT-MIC Based Matching Location Algorithm for Wire Bonding

Computer vision inspection system is often applied in automatic wire bonding to ensure the location accuracy and speed, while feature selection affects the performance of inspection system. In this paper, a new feature extraction and matching location algorithm based on SIFT-MIC is proposed for wire bonding. The algorithm chooses MIC corners in the neighborhood of SIFT feature points to increase the robustness of feature extraction. Meanwhile, the selection of SIFT- MIC feature points is before the generation of SIFT descriptors, to reduce the computation and improve the efficiency of feature extraction. The vector angle method is chosen to finish the matching location of the eye-point images and images to be matched. Experimental results show that SIFT-MIC algorithm is invariant to translation and rotation, and has high performance in the positioning accuracy and detection efficiency. The proposed algorithm is suitable for MEMS wire bonding images with rich points and lines features.