Block Sparsity Based Chirp Transform for Modeling Marine Mammal Whistle Calls

Abstract

In this paper, we propose a block sparsity-based method to reconstruct marine mammal whistle calls. Considering the block structure of harmonic chirp, we form an overcomplete group chirp dictionary to enhance the estimating resolution under iterative greedy approach. The resulting sparse estimator, also named Block Orthogonal Matching Pursuit based Discrete Chirp Fourier Transform (Block-OMP based DCFT), is able to achieve an acceptable estimating result in a noisy environment. We illustrate the performance of the resulting estimator using both simulated and collected dolphin whistles.