Spectrogram Enhancement By Edge Detection Approach Applied To Bioacoustics Calls Classification

Accurate recognition of sound patterns in spectrograms is important step for further recognition applications. However, background noise forms fundamental problem regardless the species under study. In this paper, crest factor feature was extracted from the limited dynamic range spectrogram. The developed crest factor image behaved as smoothed version of the spectrogram, at which edges of the involved sound patterns were detected without the need of prior smoothing filters and their scaling constraints. Attached noise – surrounds the detected edges – was removed, to form the enhanced spectrogram. The method was compared to other enhancement approaches such like spectral Subtraction and wavelet packet decomposition. Comparison was performed on different structure patterns of bats and birds. Results indicate how the method is promising for efficiently enhancing the spectrogram while preserving its temporal and spectral accuracy. The method correctly classified three bioacoustics species with an accuracy of 94.59%, using few 2D features of their enhanced spectrograms .