Digital Signal Processing

Gladrene Sheena Basil ¹, Juliet Vanitharani ², K. Jayapriya ³, A. Antonysamy ⁴

Affiliations
1 Mother Teresa Women’s University, Kodaikanal, Tamil Nadu, IN
2 Bat Research Laboratory, Department of Zoology and Research Centre, Sarah Tucker College (Autonomous), Tirunelveli-627007, Tamil Nadu, IN
3 Vin Solutions, Plot No.40, Raja Rajeswari Nagar, Tirunelveli– 627007, Tamil Nadu, IN
4 Anandha College, Devakottai-630 303, Sivagangai District, Tamil Nadu, IN

Abstract

Bats are unique among mammals which are key informers of climate change and play a vital role in maintaining eco-balance and pest-management in a forest, apart from being the best bio-indicators. Insectivorous bats make use of Echolocation Calls which are species-specific. Hence Echolocation calls can be used for species classification of bats. The echolocation calls are recorded using bat detectors. The audio signal is pre-emphasized and pre-processed and then features are extracted. In this paper, Discrete Tchebichef Transform and Cepstral features are made use of, to build a bat classification scheme for bats in Kalakad Mundanthurai Tiger Reserve. On classifying the bats calls with k-NN, Bayes and SVM classifiers, SVM was found to outperform others.

Keywords

Insectivorous Bats, Echolocation Call, Feature Extraction, Cepstral Features.

Full Text