Multi Feature Analysis of Smoke in YUV Color Space for Early Forest Fire Detection

Abstract

An image processing approach for detection of smoke in video using multiple features is proposed in this paper. It is assumed that the camera monitoring the scene is stationary. Video smoke detection methods have many advantages over traditional smoke detection methods due to large coverage area, fast response and non-contact. In order to reduce a false alarm rate, we propose a novel method to detect smoke by analyzing its multiple features. It consists of three stages. In the first stage, color filtering is performed in YUV color space to segment the candidate smoke region. In the second stage, spatio temporal and dynamic texture analysis is performed on the candidate smoke region to extract the spatial and temporal features; these features include wavelet energy, correlation and contrast of smoke. In the third stage, the extracted features are used as input feature vectors to train the Support Vector Machine (SVM) classifier, which is used to make decision about candidate smoke region. The proposed algorithm has been tested using news channel videos and videos captured by surveillance CCTV camera and shows impressive results in terms of detection accuracy, error rate and processing time.