Computer Vision Techniques for Firebrand Detection and Characterization

One of the most important mechanisms for fire spread in the wildland urban interface (WUI) is the transport of firebrands. These small burning particles have been found to be responsible for more than half of home ignitions in WUI areas, however techniques to characterize their production and transport are limited. This study integrates various computer vision techniques to develop a novel particle tracking algorithm than can be used to quantify exposure from firebrand showers. Application of the algorithm to video from a full-scale burning test of a Leyland cypress tree through a wind field delivers promising results for the calculation of the number, surface area, and mass flux of generated firebrands.