Edge information based crater detection and matching for lunar exploration

The precision of lunar soft landing could be boosted largely when craters are used as navigational landmarks since their distinct features. Aiming at the navigational application, this paper presents an edge information based method of crater detection and matching. In detection algorithm, considering the particularity of edges, multi-scale adaptive Gaussian filtering and adaptive Canny algorithm using histogram of image are introduced into edge detection, then craters are extracted by steps of edge selection, edge pairing and ellipse fitting. Crater matching consists of the matching algorithm based on Conic-pair Affine Invariants, and the correction algorithm for false-match and mismatch situations using affine properties of close curves' area. Experiments of real lunar images show that the detection algorithm is provided with adaptive ability to different illuminations and terrains, and the matching rate of crater matching algorithm is better than 85% with less than 5% false-match.