Spherical coordinates framed RGB color space dichromatic reflection model based image segmentation: Application to wildland fires' outlines extraction

Abstract

Wildland fires represent a major risk for many countries over the world. For efficient fire fighting, the modeling and prediction of fire front propagation is a curial need. However, wildland fires' involves complex dynamics and mathematical modelling of such complex systems needs reliable information extraction from real situations, which is far from being a trivial task. Artificial Vision and Image Processing offer appealing potential toward reliable extraction of required information. In this paper we focus on flames' and fires' segmentation, dealing with the above-stated already open problem. The segmentation approach that we propose is based on dichromatic reflection model reformulated on a spherical interpretation of the RGB color space.