Bat Classification based on Discrete Tchebichef Transform and Cepstrum, in Kalakad Mundanthurai Tiger Reserve

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.