Application of shearlet transform in the star extraction

In order to realize high precision and high accuracy of star pattern recognition and attitude determination, the star extraction algorithm is studied and the Shearlet transform is applied to the star extraction technology. First, Shearlet transform is used for the decomposition of star image to get coefficients of different scales, different directions. Then threshold processing is used to shear coefficients and reconstruct them to get the de-noising image. Then Top-hat transform and adaptive threshold processing are used for the reconstruction image to complete the star image filtering. Finally the centroid error compensation method is used for extraction of star coordinates, which effectively completes the star extraction. The experimental results show that the noise removal is very effective with star image filtering based on shearlet transform. The error by the centroid error compensation method is about 0.003 and the method is obviously superior to the traditional centroid method. It can satisfy the star sensor requirements of high precision and strong anti-jamming.