Bernoulli Track-before-detect Algorithm for Distributed Target with Unknown Amplitude Information

Track-before-detect algorithm is an effective solution of detecting and tracking a low signal-to-noise ratio (SNR) target whose amplitude distribution characteristic is needed as the prior information. Under the point target hypothesis in most previous works, the amplitude of target is usually assumed to be known or modeled as a state variable. However, these approaches cannot simply be migrated to the extended target tracking problem because of the unknown target extended length. This paper considers the issue of extended target joint detection and tracking with unknown amplitude distribution information through the use of Bernoulli particle filter based track-before-detect (BPF-TBD) methods. The proposed heuristic algorithm accumulates the multi-frame measurement data along the potential track of the target by a sliding window and then extracts the amplitude distribution information by means of principal component analysis (PCA) method. Simulation results show that the property of the proposed method asymptotically converges to the exact filter with prior correct expected amplitude distribution information, which indicates a superior performance in terms of feasibility and effectiveness.