星图像的亚像元细分定位方法研究

摘要

The common methods for sub-pixel location is introduced, i.e. centroid algorithm, Gaussian fitness algorithm, paraboloid fitness algorithm. The Gaussian fitness algorithm has the highest precision in theory, however when experiment is undertaken on real star image shot in our project the resuh is far from ideal and the error goes beyond the permitted range. Thus eentroid algorithm is taken to calculate the coordinates of stars in the image. This time experiment proved that the method's precision completely meet the need of sub-pixel level and the application for later moment.%介绍了常用的亚像元细分定位算法，即质心法、高斯曲面拟合算法和抛物面拟合算法。理论上高斯曲面拟合算法具有最好的精度，然而通过对实际星图的实验，发现其与实际情况符合的并不好，定位误差超出了允许的范围。因此本文选用质心法计算星体的坐标，并经过仿真实验证明了质心算法达到了亚像元的精度，完全满足后续应用的要求。