Abstract: Traditional power spectrum detector is based on the assumption that mean and variance of background noise spectrum do not vary with frequency. However, actual noise spectrums generally do not meet the hypothesis in non-cooperative satellite communication environment, so the detector performance is restricted. In this problem, received signals are preprocessed by sliding widow least square to make noise spectrums satisfy the presumption. Consequently, a blind detection algorithm of satellite communication signals is presented, and improving factor of detection performance relative to power spectrum detector is also deduced. Simulation results indicate that under same conditions, the algorithm generally has better performance than power spectrum detector, which is a classical traditional blind detection method. The proposed algorithm is also easy to implement with low computation complexity.