Stereovision based algorithm for measuring fire spreading characteristics

This paper presents a new algorithm for automatically measuring fire spreading characteristics by means of a stereovision system. The proposed framework is based on the use of a pre-calibrated trinocular stereo camera. The first and second steps permit the segmentation and features detection in the images. The third step is a stereo matching approach used in order to automatically process successive images of spreading fires. In the fourth step, stereovision techniques are applied in order to derive the 3D fire structure. In the last step, new developed algorithms are used to estimate, in the case of complex fire spreading, information like the position of the fire over time, its rate of spread and its height.