Ship Detection in Radar Image Series Based on the Long Short-Term Memory Network

摘要

Ship detection is one of the important ocean applications of radar images. However, researches for ship detection in low-resolution images are relatively scarce. To improve the ship detection performance in low-resolution conditions, the paper adopts the range-Doppler (RD) images and takes advantage of the multi-frame information with the long short-term memory (LSTM) network. In this paper, the interpolation method and the LSTM method are proposed, which have the advantages of speed and precision respectively and show strong anti-interference ability.