摘要
采用多边形简化的方法提取出包含源图形主要特征点的多边形.在简化多边形的类正切空间表示下,利用图形对应边在渐变过程中所掠过面积总和最小这一特征构造相似度量函数,由动态规划算法求解实现初始和目标简化多边形之间的顶点对应,再进一步建立源图形顶点之间的整体对应,最后通过插值边和角的方法实现图形渐变.实验结果表明:该算法简单有效,对应效果自然、合理.
This paper presents an effective approach for establishing vertex correspondence between two planar shapes. Correspondences were first established between the major feature points extracted from both simplified polygons of source and target shapes. We used the representation of the simplified polygons in the similar tangent space to construct a new similarity metric function by the fact that the corresponding edges transformed to each other with the minimum total amount of moving area. The resulting solution tended to associate regions on the two shapes which looked alike. Then the optimal correspondence was obtained by an efficient dynamic programming technique, meanwhile, we could attain the shape blending between the planar shapes using intrinsic method. Experimental results show that our method is fast, robust, and feature preserving.
出处
《计算机辅助设计与图形学学报》
EI
CSCD
北大核心
2007年第3期304-310,共7页
Journal of Computer-Aided Design & Computer Graphics
关键词
简化多边形
主特征点
类正切空间
相似度量函数
动态规划图
simplified polygon
major feature point
similar tangent space
similarity metric function
dynamic programming (DP) graph