Abstract: To process the low-SNR echo signal of active sonar, a method of compressed sensing based on frequency prior information (CSFPI) is proposed in this paper, which combines the prior information of incident signal and target echo signal of active sonar with compressed sensing theory. The prior information of the incident signal of active sonar in frequency domain is obtained to incorporate into the sparse decomposition process as an ‘atom’ for the construction of a complete prior sparse matrix. The echo signal of active sonar is equivalent to the linear representation of the incident signal carrying target information. By combining the feature of the echo with the orthogonal matching pursuit (OMP) algorithm, a "block" orthogonal matching signal reconstruction algorithm is formed. First, the validity of CSFPI method is verified by simulation. Then, the CSFPI method is used to process the measured data in lake trial. When the lake date compression ratio is 50%, and the signal to noise ratio (SNR) is low to -5 dB, the matching rate of reconstructed signal is higher than 70%. The experimental results show that the CSFPI algorithm is effective in dealing with sonar signals of low SNR.