一种基于自适应模糊线段的离散曲率估计

Discrete Curvature Estimation Based on Adaptive Fuzzy Segments

针对Z2空间中8-连通离散曲线的噪声影响,提出“序”为d的模糊线段生长算法.将曲线点上生长出的最长模糊线段作为切线的近似,并根据曲线局部粗糙度自适应地选择序,在此基础上进行离散曲率估计.实验结果表明:通过自适应选择序值,最长离散模糊线段不仅较好地反映了曲线点的局部特性,而且增加了对离散曲线噪声的适应能力,离散曲率估计的性能明显提高.

This paper proposes a practical algorithm for estimating discrete curvature of 8-connected curve in 2D space based on adaptive fuzzy segments. The algorithm estimates digital curvature based on tangent orientation, where tangents are approximated by the longest fuzzy segments grown from points on the curve with a given order of d. The adaptive choice of order d at each point according to the local curve coarse degree makes our algorithm particularly suitable for noisy curves. Experimental results show that our algorithm can substantially improve the performance of curvature estimation in keypoints＇ detection, and the obtained results manifest a better consistency with the features in continuous space.