1136

To solve the low efficiency, low precision, and high cost of detection in dispensing manufacture, a novel method based on machine vision combined with the optimized iterative closest point(ICP) algorithm is proposed in this paper. According to the characteristics of the dispensing products, Hough transform is selected to extract geometric features of glue points. Then, we choose template matching as coarse matching before ICP matching and find the alignment points for ICP matching to improve the efficiency and accuracy of ICP algorithm. The results of the experiment indicate that this method can accurately detect position deviation and geometric defects and its detection rate of products reaches 91.8%.