Typical existing fire detection algorithms for airborne and satellite based imagers employ the Planckian radiation in the 3.5 -5 μm and 8 - 14 μm spectral regions. These algorithms can have high false alarm rates and furthermore, the issue of validation of subpixel detection is a lingering problem. We present an empirical testing of fire detection algorithms for controlled and uniform burning and hot targets of known area. Image data sets of the targets were captured at different altitudes with the Modular Imaging Spectrometer Instrument (MISI). MISI captures hyperspectral
VNIR and multispectral SWIR/MWIR/LWIR imagery. The known range of target areas ranges from larger than the MISI IFOV to less than 0.5% of the IFOV. The in situ temperatures were monitored with thermocouples and pyrometers. Spectroradiometric data of targets and backgrounds were also collected during the experiment. The data were analysed using existing algorithms as well as novel approaches. The algorithms are compared by determining the minimum resolvable
fire pixel fraction.
Fire detection has been an active research field for many years and a number of algorithms have been proposed. These algorithms, however, are often inflexible in dealing with the spatial and temporal heterogeneity of the environment. Different biomes, seasons, and temperatures usually cause the performance of these algorithms to vary dramatically. In this paper, we propose a new algorithm for fire detection based on the Mahalanobis distance that exploits the statistical properties of multi-spectral images. The distinguishing feature of our algorithm is its robustness. It can effectively differentiate fire from background in various environments, using a single, fixed threshold. We evaluate our algorithm by comparing it to three state-of-the-art existing algorithms: the MODVOLC normalized fire index algorithm, the Arino's threshold algorithm, and the contextual MODIS algorithm. All algorithms are tested using MODIS images taken in different parts of the world as well as at different times. Experimental results demonstrate that our algorithm consistently achieves the best performance, showing a low and constant false alarm rate.