A Feature Selection Method for Craters in Martian Terrain Matching

Mars spacecraft facing severe Mars geographical environment in the landing process.Performance of feature extraction and feature matching have been seriously affected.To solve this problem, a feature selection method based on multi-feature fusion is proposed.Firstly, the AKAZE feature detector with the best comprehensive performance is obtained by using the repeatability evaluation of the feature pits.Then, based on the AKAZE feature detector, the recall rate 1-precision curve is used to evaluate the feature descriptors.Finally, one or several features with better performance are combined to extract and match the features of the craters, and the best matching performance is selected as the feature of positioning for crater.Experimental results show that compared with the traditional SIFT feature matching, the matching accuracy is improved by about 8%.