The Multicriteria Constrained Stochastic Matched Filter For Underwater Bioacoustic Signals

Bernard Xerri (1) , Bruno Borloz (1) , Maissa Chagmani (1)

1 IM2NP - Institut des Matériaux, de Microélectronique et des Nanosciences de Provence

The aim of this paper is the detection of a bioacoustic signal embedded in several noises such as sea noise and other bioacoustic signals (dolphins, sperm whales). All the signals are real world signals. Only second order statistics are use through the estimated correlation matrices of the signals. This paper proposes an extension of the Constrained Stochastic Matched Filter (CSMF) based on the optimization of the Signal to Noise Ratio after linear filtering. The approach proposed is a multicriteria one, merging three different versions of the CSMF, and is named Multicriteria CSMF (MCSMF). The objective is that the results obtained are better than the other methods, or at least equal to the best among the three. The results are provided on ROC curves and the method is compared to the classical method Stochastic Matched Filter (SMF).

Mots clés: detection; bioacoustics; Signal to Noise Ratio (SNR); stochastic matched filter; multi-noise; correlation matrix

Domaines: Traitement du signal et de l'image [eess.SP]

Bernard Xerri, Bruno Borloz, Maissa Chagmani. The Multicriteria Constrained Stochastic Matched Filter For Underwater Bioacoustic Signals. Oceans 2019 (Marseille, France), MTS/IEEE, Jun 2019, Marseille, France. ⟨hal-02178238⟩