Automatic target recognition based on receiver optimization of photonic time-stretched coherent radar.

ABSTRACT

We demonstrate an automatic target recognition (ATR) scheme based on an improved photonic time-stretched coherent radar (PTS-CR). The reception apertures of the PTS-CR can cover the entire detection range by receiving the echo signal with high repetition rate pulses and increasing the amount of dispersion of the first dispersive medium in the receiver. Two channels with different stretching factors are simultaneously used to restore the signal delay information. Simulated and experimental results verify the feasibility of the new scheme. Finally, based on the improved receiving scheme, PTS-CR successfully performed ATR on four different targets placed on a rotating stage. Combining this with the training of the convolutional neural network (CNN), the recognition accuracy rate is 94.375%.