Automatic recognition of radar signals based on time-frequency image character

Radar signal recognition is one of the key technologies of modern electronic surveillance systems. Time-frequency image provides a new way for recognizing the radar signals. In this paper, a series of image processing methods containing image enhancement, image threshold binarization and mathematical morphology is utilized to extract the shape character of smoothed pseudo wigner-ville time-frequency distribution of radar signals, and then realize the identification of radar signals by the character. Simulation results of eight kinds of typical radar signals demonstrate that when signal noise ratio (SNR) is greater than -3dB, the Legendre moments shape character of the time-frequency image is very stable. Moreover, the recognition rate by the character is more than 90% except for the FRANK code signal when SNR ≥ -3dB. Compared with existing algorithms, the proposed algorithm also has advantages of less character dimension and shorter computation time.