Auto and parallel disposing of feature points extracting and matching of orbits of Chang´e 2 selenograph

The auto-process of extracting feature points and matching them of selenograph with good topology and precision is an important, fundamental and challengeable work. Based on the angles between the front and back camera lens and the ephemeris of the Chang´e 2 satellite, lots of control points are extracted and matched automatically of the same orbit. And then, pieces of the selenograph are resampled by the extensive epipolar geometry, then the quasi-dense matching is implemented by the multi-layered image matching algorithm quickly. Thus, we can determine the probable relationships of the neighbor orbits´ selenograph, and the matching between the neighbors can also be resolved, which helps the further mosaic and three dimensional calculations. At last, the whole process is executed parallelized by our parallel algorithms for the steps of feature points extracting, matching and so on. After massive testing on many orbits of the selenograph, ours run normally and robustly with sub-pixel precision and good topology.