A novel linear time corner detection algorithm

Abstract

In areas like pattern recognition, motion analysis, outline capturing, reconstruction of objects etc, the corners of an object play a very vital role as of features for shape representation and analysis. We have presented a novel scheme for detecting corners of a planner object SRM05. The core of the algorithm is based upon slope analysis. The complexity of our algorithm is linear. It is very efficient and as well as accurate. This scheme works for both, smooth planner and irregular planner shapes. Test results of this algorithm are compared with some commonly referred corner detectors. The experiments show that this method leads to good quality results and robustness to noise with low computational cost.