Radar Signal Separation Recognition Method based on Semantic Segmentation

With the increasing application of electronic technology in military field, electronic countermeasure technology has been developed gradually. The separation and identification of radar signal is an important part of electronic countermeasure. Only when the enemy's information is fully grasped in the war, can the absolute advantage be obtained in the battle, which makes the separation and identification of radar signal play a very important role. However, radar signal identification is faced with serious time-frequency domain overlap problem, and the analysis of multi-component radar signals and the acquisition of valuable information are still faced with great difficulties, which is an urgent problem to be solved in radar reconnaissance system. In this paper, a one-dimensional signal is represented as a time-frequency graph (TFIS) by using the Choi-Williams distribution (CWD), and then the recognition results are obtained by using UNet to segment the time-frequency image of the signal, and the recognition results are visualized. The simulation results show that the proposed method can effectively solve the multi-signal separation and recognition of time-frequency aliasing.