3D Reconstruction Of Lunar Surface Based On Structured Feature-Matching

For the navigation of Lunar Exploration Rover (or Lunar Rover Vehicle, LRV), localization is one of the most important problem. The localization scheme based on matching between the global and the local lunar maps is of great importance for the global localization. In the map-matching based localization scheme, one of the key problem is the mapping according to the lunar images acquired both on-orbit and on-vehicle. In this paper, aimed at the map-matching based global localization of LRV, digital lunar map reconstruction techniques are studied, some methods are proposed and implemented.In this paper, based on the matching among the parameterized contour feature, a 3D lunar surface reconstruction method is proposed. The reconstruction method is a stereo vision-based modeling method, in which the contour feature of the characterized object is extracted from the lunar surface stereo-images, and the objects'geometric information is inferred according to the correspondence between the parameterized contour features, therefore, the digital lunar map is constructed. Using the crater as the instance, the discussions are focused on the following topics: contour feature extraction, structured feature recognition, structured feature parameter reconstruction, and parameterized contour feature matching. A region contour feature extraction method is proposed, which is based on clustering and linear determination, noises are removed with a mathematical morphology process, and the experiment result shows that the method can extract the crater's contour effectively. On observation of the craters'characteristic, a feature relation based target recognition method is proposed, and contours are fitted by parameterized curves, the geometric information of the structured features is recovered through stereo-vision principle. The method is verified by the experiment result in the reconstruction of craters. A hierarchical structured feature matching method is proposed, which is utilized in the matching between the global map and the local map, the method is implemented and verified by the experiment result.In this paper, the images from the Apollo 15 program are used as reference data. The methods proposed in this paper are all implemented through prototype method. The experiment result shows that the methods based on parameterized contour feature matching can reconstruct the 3D lunar surface effectively...