Paper Title
Smoke Detection Approach Using Wavelet Energy and Gabor Directional Orientations
Abstract
Sensing smoke at early stages is vital for averting fire events. This study proposed a video based approach that can
detect smoke based on computations made from wavelet energy and directional orientations obtained from Gabor filter banks
considering the characteristics such as diffusion, color, semi-transparent property. In the proposed model, pixels of smoke
colored are identified by masking in HSV color space and a frame difference is applied to detect the motion. To extract the
temporal feature vectors, we proposed a new method that determines the texture of smoke by applying Gabor filter bank with
preferred orientations. In addition, wavelet energy is computed and is fed as another feature in to feature vectors. Finally these
features are fed to a support vector machine (SVM) to discriminate our data more thoroughly and provide accurate detection
of smoke. Experiments are carried out with benchmark datasets showing the proposed approach can work effectively with
non-false alarm.
Index Terms- Gabor Filter, Smoke Detection, Support Vector Machines, Temporal Features.