基于B-Snake的三维图像骨架化方法  

3D Image Skeletonization Based on B-Snake

作　　者：田绪红[1,2] 韩国强[2] 司徒志远[1] 陈茂资[1] 宋鸿陟[1] 

机构地区：[1]华南农业大学信息学院,广州510642 [2]华南理工大学计算机科学与工程学院,广州510642

出　　处：《计算机科学》2008年第9期241-244,265,共5页Computer Science

基　　金：国家自然科学基金项目(60573019);广东省自然科学基金重点项目(05103541)

摘　　要：3D图像数据由于其本身的复杂性,对其进行骨架化一直是3D图像处理中的一个难点问题。文中提出了一种B-Snake骨架化算法及B-Snake能量最小化求解法,同时讨论一种骨架化算法的评价方法。B-Snake骨架化算法获取的骨架其连通性和光滑性隐含于B-Spline表达式中,而且在中心性及抗噪声性等方面也明显优于离散Snake算法。对模拟的测试数据和实际三维重建的植物根系图像进行了实验,结果都表明了文中提出的算法的有效性。Skeletonization of 3D image is one of hard problem in 3D image processing. A B-Snake based algorithm to extract skeleton from 3D image and a solution to energy minimization of B-Snake are proposed in this paper. An evaluation method of 3D skeletonization algorithm is also discussed. The skeleton extracted by the proposed algorithm has nice connectivity and smoothness as it is implicit in the B-Spline representation. It also has better medialness, and less sensitivity to noise compared with Discrete Snake algorithm. Experiments on simulated testing images and real plant root images confirm the proposed approach.

关 键 词：三维图像骨架化 离散Snake B-SNAKE 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] U495[自动化与计算机技术—计算机科学与技术]