点击次数：

论文类型：期刊论文

发表时间：2022-06-28

发表刊物：自动化学报

卷号：42

期号：7

页面范围：1100-1112

ISSN号：0254-4156

摘要：A fast skeleton extraction algorithm is proposed in this paper. Firstly,
   some vertices of the polygon contour are labeled as salient convex
   vertices by using discrete curve evolution, then the polygon contour is
   partitioned into a main branch and several horizontal branches according
   to the types of these salient convex vertices. Secondly, the horizontal
   secant lines method is used to get the skeleton of the main branch and
   the vertical secant line method is used to obtain the skeletons of
   horizontal branches separately. Finally, by connecting the skeletons of
   horizontal branches to main branch skeleton, the final skeleton of a
   given contour is obtained. Experimental results show that the proposed
   skeleton extraction algorithm has good performance on Kimia data set. In
   addition, results on pedestrian video are presented to show that the
   proposed algorithm performs well on natural images, too. Furthermore,
   the time complexity of the algorithm is lower than those of other
   classical algorithms and it satisfies the requirement of real-time
   processing of videos.

备注：新增回溯数据