An active detection method for an underwater intruder using the alternating direction method of multipliers

The underwater surveillance of harbors is challenging, as the reverberation-limited environment means that intruders with a weak target strength, such as autonomous underwater vehicles and frogmen, may be overlapped or even overwhelmed by reverberations. In recent years, developments in low-rank and sparsity theory have presented an innovative idea for intruder detection. This paper describes a method in which multi-frame data containing echoes of an intruder and reverberations are arranged in a matrix, and then the data matrix is decomposed to achieve detection. Based on this idea, the alternating direction method of multipliers (ADMM) is applied to the problem of intruder detection. The key parameters of ADMM are discussed and recommended values are given. The proposed method is validated and evaluated through a series of experiments.