Monitoring and Extracting Abnormalities in Land Surface Temperature Images for Automatic Identification of Forest Fires

Forest fires have a detrimental impact on economy and environment. The rapid distribution of the fire could cause many causalities and a lot of effort is required to control. To overcome this problem it is highly important to detect forest fire before it spread its wing on the surroundings. A number of popular techniques like smoke velocity distribution, usage of sensors and sounding systems have been applied in forest surveillance and found to be ineffective. In this paper, we present improved forest fire identification through the real time processing using land surface temperature satellite imagery. From these images an analysis is carried out to identify the mean wavelengths of abnormal temperature distribution when compared to the surroundings on a small region. If the mean wavelength exceeds 10.14, it is treated as forest fire. This approach uses k-mean clustering and haar wavelet, resulting an average accuracy rate of 89.5 %.