Frequency line extraction on low SNR lofargram using principal component analysis

Frequency line tracking in a lofragram is usually challenged by the strong background noise in passive sonar systems. This paper proposes a principle component analysis based lofargram analysis method for multiple frequency line tracking. The simulation and experiment are conducted by a set of simulation data and a set of ship noise data recorded on the South Sea of China, respectively. The results demonstrate that our method can provide more spectrum information than the reference methods by enhancing the weak frequency lines seriously interrupted by the background noise.