To overcome the disadvantages of a non-cooperative frequency hopping communication system, such as a high sampling rate and inadequate prior information, parameter estimation based on Blind Compressed Sensing (BCS) is proposed. The signal is precisely reconstructed by the alternating iteration of sparse coding and basis updating, and the hopping frequencies are directly estimated based on the results. Compared with conventional compressive sensing, blind compressed sensing does not require prior information of the frequency hopping signals; hence, it offers an effective solution to the inadequate prior information problem. In the proposed method, the signal is first modeled and then reconstructed by Orthonormal Block Diagonal Blind Compressed Sensing (OBD-BCS), and the hopping frequencies and hop period are finally estimated. The simulation results suggest that the proposed method can reconstruct and estimate the parameters of noncooperative frequency hopping signals with a low signal-to-noise ratio.