Star sensors may work in star tracking cases and predictive centroiding algorithm cases when it has sufficient attitude knowledge,which not only can increase the update rates of star sensors,but also can improve recognition performance.A novel star tracking algorithm for a star sensor with a large angle manoeuver is presented to avoid disadvantages of classic star tracking algorithms.In order to decrease noise errors of the star sensor,the filtering algorithms for star images and star locations are presented.The experimental results show that the star locations are obtained in the threshold scan window 10 pixel×10 pixel in the star image and the attitude accuracy of the star sensor is improved by 5″ in the star image filtered by C-L wavelet,when the spacecrafts are rolled with a angle velocity of 2.25°/s.These algorithms are tested at night sky experiments in 2007 and in 2008 and will be used in the attitude control system of a satellite in 2009.