Real-time fire segmentation via Active Contours for UAV integrated wildfire propagation prediction
- 7 January 2018
- conference paper
- conference paper
- Published by American Institute of Aeronautics and Astronautics (AIAA) in 2018 AIAA Information Systems-AIAA Infotech @ Aerospace
Abstract
Accurate wildfire spread prediction is a key element in planning effective ground and aerial operations. Because of the underlying complex dynamic multi-physics processes driving the forest fire phenomena and the high number of parameters involved, finding an analytical solution is a challenging task. Current operational wildfire spread simulators, used by national governmental agencies are FARSITE, PROMETHEUS, PHOENIX RapidFire. These tools are based on empirical models developed and tuned using laboratory and historical wildfire data. This aspect makes the solution provided by these simulators inaccurate over long periods of time. To overcome these limitations, a closed loop architecture, where real time field measurements are fed back into the system, is the most promising solution. In this scenario, the use of an unmanned platform considerably reduces the risk associated with flying a manned aircraft in a low visibility and extremely turbulent air and improves the on-board Electro- Optical (EO) sensor effectiveness by flying at very low altitudes. In this paper a robust fire segmentation algorithm for wildfire front tracking is presented. This algorithm is based on the solution of Partial Differential Equations (PDE) to model a time evolving curve. An efficient implementation of the Level Set method enables the algorithm to fulfil real time requirements. Flight tests over a prescribed burn have been carried out to collect real data about the fire dynamics and to validate the algorithm and to test its robustnessKeywords
This publication has 33 references indexed in Scilit:
- Overview of Forest Fire Propagation ResearchFire Safety Science, 2011
- Wildland surface fire spread modelling, 1990 - 2007. 2: Empirical and quasi-empirical modelsInternational Journal of Wildland Fire, 2009
- Active contours without edgesIEEE Transactions on Image Processing, 2001
- Prediction of diurnal change in 10-h fuel stick moisture contentCanadian Journal of Forest Research, 2000
- Current approaches to modelling the spread of wildland fire: a reviewProgress in Physical Geography: Earth and Environment, 1998
- A Fast Level Set Method for Propagating InterfacesJournal of Computational Physics, 1995
- Prediction of crown fire behavior in two stands of jack pineCanadian Journal of Forest Research, 1993
- Fronts propagating with curvature-dependent speed: Algorithms based on Hamilton-Jacobi formulationsJournal of Computational Physics, 1988
- Fire Temperatures in Grass, Shrub and Aspen Forest Communities of Central AlbertaJournal of Range Management, 1980
- Conditions for the start and spread of crown fireCanadian Journal of Forest Research, 1977