Recurrent neural network for high-resolution radar ship target recognition
Wang Feixue, YU Wen-xian, Guo Guirong · 2002
The high-resolution radar waveform describes the amplitude of targets' multiple scattering centers and their distribution in the radial axis. As viewed from the time domain, the target waveform can also be regarded as a time sequence such that it can be classified using recurrent neural networks (RNN) which are suitable for time sequence processing. A modified partially RNN and its algorithm are proposed. This method reaches an average recognition rate of above 90% for 8 class high-resolution radar targets, and it is tolerant of time shift to a certain degree.