IMF-Based Energy Spectrum Distribution Permutation Entropy Signal Detection Method for the Marine Mammals Sounds

Passive acoustic monitoring (PAM) systems play a crucial role in studying marine mammal sounds and their ecological significance in complex underwater environments. This paper introduces an innovative approach, the intrinsic mode function energy spectrum distribution permutation entropy (IMFi-ESDPE) algorithm for detecting cetacean sounds. The algorithm combines empirical mode decomposition (EMD) for signal analysis, energy characterization, and information theory entropy for feature extraction and detection. The study is divided into two main components: signal sampling and feature extraction and detection algorithm. In the signal sampling phase, cetacean sounds are collected and divided into equal-duration sampling signals. The feature extraction and detection algorithm utilizes EMD to decompose the signals into intrinsic mode functions (IMFs), calculates the energy distribution and average energy ratios of each IMF, and computes the Hilbert energy spectrum for comprehensive representation. Permutation entropy is then applied to the energy distribution function of each IMF, resulting in the IMFi-ESDPE values. By setting optimized thresholds based on the distribution of permutation entropy, the algorithm achieves high accuracy in detecting cetacean sounds. The proposed method holds great potential for application in marine mammal sound signal detection, contributing to passive acoustic monitoring systems and enhancing our understanding of marine ecosystems and conservation efforts.