A Novel Algorithm for Detecting both the Internal and External Corners of Checkerboard Image

Aiming at the limitation of current algorithms for corner detection, a novel algorithm is proposed for detecting both the internal and external corners of checkerboard image in this paper. Firstly, initial locations of corners are obtained by introducing perspective projection transform. Secondly, within the area of initial locations, internal corners of checkerboard image are detected with symmetry and variance operator. Based on the univalue segment assimilating nucleus idea in algorithm SUSAN, quarter operator which can detect external corners is defined. Finally, the sub-pixel locations are achieved through a polynomial approximation of the corner response function. Experimental results show that the proposed algorithm is feasible and effective.