An effective method for corner detection

The conventional Harris algorithm for corner detection, though with high reliability and strong stability, is unable to meet the requirement of the real-time system. Focus on the problem of the great amount of calculating when extracting corners on the image, this paper proposes a reduction algorithm based on Harris algorithm as follows. In order to reduce the amount of calculating and improve real-time performance of the system, we filter the pixels across the image before other steps, that is establish a threshold, exclude the pixels have similar gray value with the target pixel in its 3 * 3 field before other steps. The proposed algorithm has been verified by several simulation experiments. The results of the tests indicated that the detecting time of the proposed algorithm is only about one - third of the original algorithm while the detecting efficiency is almost the same, offers a robust and effective solution for real-time processing.