On the selection of time frequency feature set to represent passive sonar data of tropical waters

The characterization of underwater ambient noise using time frequency representations (TFRs) is challenging due to the opulence of the potential information that can be extracted from it. Traditionally Short Term Fourier Transform (STFT) is used to observe underwater ambient noise sources from the signal received by sonar. Short term Fourier Transform is a class of distributions determined by window size and shape. Wavelets also have unbound applications to represent the Sonar data. In this paper, statistical and image quality parameter sets are used to represent time frequency features of littoral water ambient noise. These feature sets are compared on variance scale by applying TFRs to different combinations of natural and anthropogenic ambient noise sources recorded at the west coast of India.