基于多种修正方式的概率松弛匹配算法  

Probabilistic relaxation matching algorithm based on multi-revisory method

作　　者：潘鸿飞[1] 梁栋[1] 李薇[1] 

机构地区：[1]安徽大学计算智能与信号处理教育部重点实验室,安徽合肥230039

出　　处：《合肥工业大学学报（自然科学版）》2008年第3期462-464,共3页Journal of Hefei University of Technology：Natural Science

基　　金：国家自然科学基金资助项目(60772121)

摘　　要：文章提出了一种将谱图理论、特征点的灰度特征及空间特征和概率松弛法相结合的特征点匹配算法;首先通过谱方法求出特征点匹配的初始概率,然后利用特征点的灰度相关性及形状上下文关系来修正初始概率,再利用特征点的形状上下文关系作为初始支持度,最后将修正过的初始概率、初始支持度与概率松弛迭代法相结合,获得匹配结果;实验结果表明,该方法能够达到很高的匹配效果。A novel algorithm for point correspondence is proposed, which combines graph spectral analysis and partial characteristics of the point together via the method of probabilistic relaxation. Firstly, the initial correspondence probabilities are obtained by means of spectral analysis. Secondly, partial characteristics are employed to compute the initial support. Finally, by combining the intial probability and support with the probabilistic relaxation, the correspondence results are gained. The experimental results demonstrate that this approach can achieve comparatively high accuracy.

关 键 词：匹配 形状上下文 概率松弛 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]