Abstract: A new feature point extraction algorithm based on prior knowledge of navigation systems for lunar rover is presented in this paper .The key-points was extracted by SIFT(Scale Invariant Feature Transform) from image as feature points at first. Then use these feature points for binocular matching thus getting distance information of the scene by the calculation of differences between match points in two images. This algorithm could improve the robust to lightness change for images of lunar region and accuracy in the later matching phase. This algorithm make good results in simulative environment experiments.

Key words: lunar rover, SIFT feature, vision navigation, feature extraction, feature matching