A MEAN-FIELD ANNEALING APPROACH TO ROBUST CORNER DETECTION

This paper is an extension of our previous paper to improve the capabi lity of detecting corners, We proposed a method of boundary smoothing for curvature estimation using a constrained regularization technique in the previous paper, We propose another approach to boundary smoothi ng for curvature estimation in this paper to improve the capability of detecting corners, The method is based on a minimization strategy kno wn as mean field annealing which is a deterministic approximation to s imulated annealing, It removes the noise while preserving corners very well. Thus, we can detect corners easier and better in this approach than in the constrained regularization approach. Finally, some matchin g results based on the corners detected by corner sharpness in the mea n field annealing approach are presented as a demonstration of the pow er of the proposed algorithm.