Aimed at the poor performance of the wideband spectrum detection under low SNR, a novel cooperative wideband spectrum detection algorithm is presented based on compressed sensing for cognitive radio. By using the algorithm，first the measurements of cyclic spectrum from the signals` cross-correlation function of compressive samples are achieved on the basis of the unique sparse property of wireless communication signals in cyclic spectra region. Then all signals' cyclic spectra are reconstructed in the entire wideband by adopting the Sparse Adaptive Simultaneous Matching Pursuit (SASMP) cooperative algorithm. Simulation results show that the algorithm proposed has a better performance under Raleigh fading channel and low SNR environment, and the SASMP cooperative algorithm is improved in reconstructing quality and algorithm complexity compared with other classical algorithms.