Modified particle filter based on feature extraction coupling for track-before-detect algorithm

For traditional particle filter track-before-detect(TBD) algorithms, when tracking targets, these algorithms require inter frame accumulation of multiple frames to accurately achieve target tracking. In this paper, we modify the traditional particle filter TBD by adding the feature extraction, in which the feature extract is used to classify the target scope and clutter scope before target tracking to narrow down the scope of particle generation. The simulation results show that the modified particle filter algorithm has significantly modified tracking performance compared to traditional particle filter algorithms, and has engineering application value.