Under daylight conditions, the intense scattering of sky background light,and the star tracker works better in the near-infrared band, but due to the inherent characteristics of near-infrared sensitive devices, there are features such as salt noise, streak noise due to imagenonuniformity, backgroundnonuniformity, and strong background in the near-infrared star image. It is difficult to achieve effectively extraction of stars in a star image. In order to improve the ability of starextraction,a multiple filtering method combining contraharmonic mean filtering, energy correlation filtering and morphological filtering is proposed in this paper. It solves problems such as salt noise, streak noise, background nonuniformity and strong background. The results of simulation and experiment show that this method can effectively suppress noise and improve the quality of near-infrared star images. The error of the star extraction is within 0.5 pixel.