Enhancing the Stability of Harris-Plessey Corner Detection

W.A. Weliamto, L. Li, H.S. Seah (Singapore)

Keywords

RealTime Corner Detection, Low Pass Filter, Notch Filter,FPGA Implementation, Stability.

Abstract

Harris-Plessey is one of the most popular corner detectors recently. In this paper, we present a method for corner detection, which greatly enhances the stability of Harris Plessey method. Compared with SUSAN (Smallest Univalue Segment Assimilating Nucleus) corner detection, our proposed corner detector can detect corners in small areas accurately. The improved corner detection method utilizes Notch Filter (NF) instead of Low Pass Filter (LPF) used in Harris-Plessey corner detection. The Modified Harris-Plessey (MHP) detector is more stable and the result of corner finding is more accurate, at the same time, it retains the advantages of the Harris-Plessey (HP) method. Compared with SUSAN corner detection, which uses a large kernel, MHP can detect corners in small areas, which would be skipped by SUSAN corner detector. Another distinct advantage of the proposed method is that it can be easily implemented in hardware, and achieve the real time result in many applications. It is verified in our experiments on FPGA (Field Programmable Gate Array) implementation.

Important Links:



Go Back