MEMS chip quality detection by fast image matching method

Abstract

Chip quality detection is one of the important procedures before the picking up operation of chip by mounter. Fast image matching method was proposed to conduct the quality inspection of micro-electromechanical systems (MEMS) chip. The advantages and disadvantages of gray projection based matching algorithm were analyzed. Then a compound projective feature matching method was put forward. VC based algorithm was given to achieve compound projective feature matching method. Compared with the traditional gray matching algorithm and the normalized cross-correlation algorithm, the ability to match and the match time of the proposed matching method were studied. Contrast with single direction of projection matching method, the matching accuracy of the given matching method was tested. Meanwhile the matching ability of proposed algorithm was test for a small angle rotation of image. Finally, the compound projection-based matching method was used to detect the quality of the MEMS chip, and the practicality of the proposed method in terms of time and accuracy were verified by experimental results.