Abstract: Direction of arrival (DOA) estimation and signal recovery is the base of the underwater target localization, tracking and recognition. Based on the compressed sensing theory, a method for DOA estimation and source signal recovery is proposed using the single snapshot processing of the array data in frequency domain. The array data are transformed to frequency domain, and the single snapshot frequency domain data are regarded as the measure data of the compressed sensing. According to the frequency, searching orientation and array manifold, the overcomplete array manifold is constructed as the sensing matrix of the compressed sensing. Both the target signal and power of the searching orientation are estimated by the basis pursuit method to complete DOA estimation and signal recovery. Simulation results show that the proposed method has a number of advantages over the minimum variance distortionless response (MVDR) method, including improved robustness to noise, less requirement in sensor number and snapshot number. And the correlation coefficient of the signal reaches up to 0.89. Experiment results in real environments verify that the proposed approach performs more effectively in the detection of weak targets than the MVDR method and can be applied to real sonar system.