Abnormal activity detection in forest reserve using cumulative short time fourier transform features

Conservation of forest and biodiversity is challenging task in today's environment. Various natural calamities and illegal activities in the forest causes loss of natural resources. In this paper we proposed an automatic process to find these abnormal activities using bark band cumulative short time fourier transform. All sound signals of various activities are transformed to fourier domain using short time fourier transform, now these are converted to bark domain. Cumulative short time fourier transform of each bark band is calculated and normalized. On the calculated feature vector successful classification of abnormal activity has been done, results are more reliable to use practically to save forests.