摘要
针对传统张量投票(Tensor voting)算法计算过程复杂、算法效率低的问题,本文提出了一种二维解析张量投票算法.首先,深入分析张量投票理论的基本思想,分析传统张量投票算法的不足及其根源;其次,设计了一种二维解析棒张量投票新机制,实现了二维解析棒张量投票的直接求取;在此基础上,利用二维解析棒张量投票不依赖参考坐标系的特性,设计并求解了二维解析球张量投票表达式,解决了长期困扰张量投票理论中球张量投票无法解析求解,仅能通过迭代数值计算,计算过程复杂、算法效率低、算法精度与算法效率存在矛盾的难题.最后,通过仿真分析和对比实验验证了本文算法在精度和计算效率方面的性能均优于传统张量投票算法.
A novel 2D analytical tensor voting algorithm is proposed to reduce the complexity and heavy computational burden in traditional tensor voting. Firstly, basic thoughts of tensor voting theory are investigated, and shortcomings and corresponding reasons are analyzed. Secondly, a new voting mechanism for 2D stick tensor is proposed and an analytical solution to the proposed 2D stick tensor voting mechanism is presented. Owing to the analytical 2D stick tensor voting being independent of the particular reference coordinate system, the mechanism for2 D ball tensor voting is proposed and the analytical solution is also provided. Thus, the problems of iterated numerical approximation, complicate computational process and the confliction between accuracy and efficiency in traditional 2D tensor voting,all caused by lack of analytical solutions, are soundly solved.At last, the correctness, accuracy and efficiency of the proposed algorithm are validated through simulated analysis and comparative experimental results.
出处
《自动化学报》
EI
CSCD
北大核心
2016年第3期472-480,共9页
Acta Automatica Sinica
基金
国家自然科学基金(51305390
61501394)
河北省自然科学基金(E2012203002)资助~~
关键词
张量投票
结构推理
解析解
特征提取
Tensor voting
structure inference
analytical solution
feature extraction