Abstract：Feature based matching algorithm is important in part of image registration. In view of repeated matching, many-to-one matching and low accuracy of traditional SIFT matching method, a bidirectional feature matching algorithm based on standardized euclidean distance is proposed in this paper. Firstly, the feature points are extracted by SIFT algorithm. Then the feature descriptors are measured by normalized Euclidean Distance and the feature points are matched by bidirectional feature matching algorithm. Finally, the matching pairs are purified by RANSAC algorithm. The experimental results show that the use of standardized Euclidean Distance for bidirectional matching has higher accuracy.
2018, Vol. 1 
