The sparse reconstruction is one of the key factors in compressive sensing theory. In order to apply sparse reconstruction to compressive imaging process of star tracker in an effective way, the effect of sparse reconstruction on star image was analyzed from the perspectives of star image error and star feature error. For the problem of image error, PSNR（peak signal-to-noise ratio） evaluation index was employed to observe reconstruction quality of star image under different reconstruction algorithm and diverse compression ratio. For the problem of feature error, firstly, the influential mechanism of sparse reconstruction on star feature was analyzed theoretically; and then the evaluation index of reconstruction error of star feature was proposed to investigate the reconstruction error of star centroid, brightness and number. Result shows that, the star image can obtain higher reconstruction quality than the common image with the same reconstruction condition under diverse compression ratio of selected algorithm, and the reconstructed star can keep its features for attitude determination to a large extent. The conclusion proves the validity of attitude estimation by using the reconstructed star image, verifies the feasibility of applying compressive sensing to star tracker and provides the practical foundation for compressive imaging of star tracker.