Optimum Segmentation of Digital Planar Curves Using Quadric Bézier Curves

This paper presents a dynamic programming algorithm to detect dominant points within a planar digital curve by approximating the given curve with quadric Bezier curves. Normally, people use cubic or higher curves to piecewise approximate a given curve, and the main job is to achieve an accurate approximation instead of dominant point detection. While using cubic or higher curves in the process of detecting dominant points of a planar curve, although a higher accuracy is achieved, the computational complexity is high too. In this paper we propose an algorithm using quadric Beziers instead of circular arcs and lines to approximate the given curve. Compare with previous algorithms our method gets more reasonable results without increasing the computational complexity.