Study on the Fast Matching Location Algorithm Based on Feature Points for Wire Bonding

In order to increase the positioning accuracy and speed of PRS (pattern recognition system) of a wire bonder, a fast positioning method based on feature points was proposed and implemented in this paper. The feature points with translation and rotation invariance were extracted from both the eye point image and the test image with the SIFT (Scale Invariant Feature Transform) feature detector, and the rule of vector angle was used to extract the robust feature point pairs. Experimental results show that the algorithm can improve the speed and accuracy of the PRS and satisfy the requirement of automatic wire bonder.