Estimating the surface and volume of laboratory-scale wildfire fuel using computer vision

This work deals with fire spreading on inclined tables. The method is based on a stereovision framework based on the use of two synchronised stereovision systems. At each acquisition time, a three-dimensional shape of the fire front is reconstructed and from it the surface and the volume of the fire front are estimated. It is the first framework that is able to obtain this information from complex spreading fires. The accuracy of the system was evaluated using a physical model of a fire front and a specific shape of fuel with known position, thickness and width. The experimental results show that the obtained accuracy is within 4% using a physical model of fire.