基于距离变换的蚁群搜索骨架提取方法

发布时间：2018-01-29 21:16

本文关键词： 蚁群算法距离变换梯度多边形逼近骨架提取骨架修剪　出处：《系统仿真学报》2017年11期 　论文类型：期刊论文

【摘要】：针对基于距离变换的骨架提取方法连通性不足的问题,提出了一种基于距离变换的蚁群搜索骨架提取方法。该方法根据距离变换及其梯度获得物体的潜在骨架点;通过控制物体边界曲线的多边形逼近误差,确定了对物体形状信息贡献程度不同的边界顶点,并将其凸顶点作为物体的骨架线端点,同时根据端点内角的大小排除了冗余骨架分支的端点;将距离变换的梯度、像素间的相似性和骨架线的走向三个因素设置启发式引导函数,并提出了用路径代价值来衡量搜索路径的优劣。实验结果证明,该方法获得的骨架线不但具有良好的连通性和多尺度条件可控性,而且减少了冗余骨架分支的产生,同时还可以克服一定边界噪声的影响。
[Abstract]:The connectivity of skeleton extraction method based on distance transform is insufficient. An ant colony search skeleton extraction method based on distance transformation is proposed, in which the potential skeleton points of the object are obtained according to the distance transformation and its gradient. By controlling the polygonal approximation error of the boundary curve, the boundary vertices with different contributions to the shape information of the object are determined, and the convex vertices are regarded as the end points of the skeleton line of the object. At the same time, the endpoints of redundant skeleton branches are excluded according to the inner angle of endpoints. The gradient of distance transformation, the similarity between pixels and the direction of skeleton line are set up as heuristic guidance function, and the value of path generation is put forward to measure the merits and demerits of search path. The experimental results prove that. The skeleton line obtained by this method not only has good connectivity and multi-scale conditional controllability, but also reduces the generation of redundant skeleton branches and can also overcome the influence of certain boundary noise.
【作者单位】：南京航空航天大学机电学院;
【基金】：江苏省自然科学基金(BK20161487) 江苏省六大人才项目(GDZB-034)
【分类号】：TP18;TP391.41
【正文快照】： 引言1 骨架(Skeleton)又称“中轴”(medial axis)[1],最早是由Blum提出的,还分别给出了两种骨架典型定义:烧草模型[1]和最大球(圆)模型[2]。骨架是表示物体拓扑结构和几何形状的一种有效形式,并且自身的线形连通结构易于转化为树或图的形式,为物体的表示和识别带来了方便。

【相似文献】