Lunar Detector Landing Area Selection Technology Based On Single Image

High precision and high resolution three-dimensional stellar topography data can provideessential data guarantee for deep-space detector to land on stellar surface safely.Three-dimensional topography data generated by traditional stereo photogrammetry possesseshigh-precision and reliability, but it is difficult to obtain stereo images to generate DEM of theaim stellar surface while the landing of the Lander. If the image obtained by detector in theprocess of landing can be used, and shape from shading (SFS) technology can be adopted togenerate large-scale three-dimensional topography data by gray information of a single image,the data can provide credible support for optimal selection of landing area. SFS technology isof simple calculation and wide range application, but it is generally assumed hat the imagingconditions are ideal, so there are many practical problems need to be solved. On account of thepractical problem the thesis studied the relative technique and applied it to landing areaselection.This paper pays main attention to SFS algorithms solved by fast marching method underperspective projection, and applied it to landing area selection. The main works are as follows:Firstly, the impact on3D reconstruction caused by different noise was analyzed, the noisewas preprocessed, and the3D reconstruction precision got improved.Then, the existing classical lighting models were summed up, and a hybrid lightingmodel was proposed. The theories of every lighting model were introduced, and the fitnessbetween the models and material reflect characters were analyzed, and then a hybrid lightingmodel was proposed.Thirdly, the common methods to solve SFS were summed up, and a SFS algorithm solvedby fast marching method under perspective projection was proposed. Four classical SFSalgorithms were lucubrated, and the efficiency, advantages and disadvantages of everyalgorithm were analyzed. On account of the perspective projection, fast marching method waschosen and the shadow areas were processed. The reconstructed results with lunar imagesshow the validity of the algorithm.Finally, the SFS algorithm was applied to lunar detector landing area selection. The mainterrain and shadow characters were analyzed firstly, and the rocks and carters and nearby areaswere extrated by edge extration, then the terrain roughness is used to judge the safety of thelanding area before selecting. The selection result proved the validity of the algorithm.