A fast corner detector for fuzzy mineshaft images based on dual-threshold

We proposed a fast corner detection algorithm based on dual-threshold, which can detect the corners efficiently in real-time for the fuzzy and uneven brightness mineshaft video images. First, the detector recognizes the flat region and corner region according to four pixels in four directions around the detected pixel, and only detects the corner region to reduce the time consuming. Second, the detector marks the background and the foreground of the corner region, and then uses the separate threshold to compute the similar segment of corners (SSC). Finally, the number of the continuous pixels of the SSC is used to detect the true corners. Experiment results show that the present algorithm performs well on the fuzzy and uneven brightness mineshaft images.