A rapid defect detecting algorithm for printed matter on the assembly line

Abstract

To better adapt to defect detection requirements of the industrial printed matter on the assembly line, this paper proposed a rapid defect detection algorithm. First we improved the Surf algorithm and made better the detection of the feature point performance and speed effectively; Second, a new FIFO feature point matching algorithm was introduced that could save feature matching time about 50%; Finally we introduced a two-way image difference algorithm to avoid error detecting for contour artifact. Experimental results showed that the proposed algorithm could meet real-time requirements on assembly line.