摘要
对提出的基于马氏距离的点匹配方法进行了理论分析与实验验证,针对马氏距离及加权图转换匹配方法的不足,将马氏距离融入到加权图转换匹配算法中,提出了一种新的稳健的图像匹配策略——基于马氏距离加权图转换的图像匹配算法。该算法利用图中的点及其K-近邻点的马氏距离中值和角度距离建立权重矩阵,根据不断更新得到的权值更新图,逐个剔除出格点,获得更加精确的匹配结果。仿真数据和真实图像实验对比结果表明该方法的可行性和鲁棒性。
This paper proposed a point matching algorithm based on Mahalanobis distance and analyzed its performance. Due to the limitation of Mahalanobis distance and weighted graph transformation, it embedded the similarity evaluated by mahalanobis distance into WGTM algorithm. It proposed a novel and robust image matching strategy, the algorithm based on weighted graph transformation using Mahalanobis distance. It built the weight matrix by using the median of mahalanohis distance and angular dis- tances between edges that connect a feature point to its K-nearest neighbors in the graph. It updated the graph according to the weight ever-updating. It obtained then precise results of the point pairs matching through iteratively eliminating the outliers. Ex- perimental results on synthetic data and real-world data have demonstrated that the proposed algorithm is effective and robust.
出处
《计算机应用研究》
CSCD
北大核心
2014年第4期1256-1259,共4页
Application Research of Computers
关键词
图像匹配
马氏距离
加权图转换
K-近邻
角度距离
image matching
Mahalanobis distance
weighted graph transformation
K-nearest neighbor
angular distance
