Abstract:
Forest is considered as one of the most important and indispensable part of a
country. At least 25 of the total land area of a country should be forest in order
to maintain the perfect ecological balance. Because of the climatic condition of
that part of the world, this is a very common scenario in those countries and
a constant threat. This happens randomly throughout the whole year. Still no
one has invented any method to detect or predict forest fire before its occurrence.
So we can’t stop it from happening when it occurs due natural causes.
But if early detection can be made, we can save a lot of wild lives as well as we
can get rid of mass destruction. Our main target there is to detect the fire at a
very early stage to take proper measures to stop it from spreading. In this paper
we used videos of fire regions. But the detection method is image based. We
segmented the video into images and detected fire from those images based
on some definite experimental rules. Finally the decision is made based on
the threshold values of fire pixels. We went through several filter threshold
values in order to minimize the false alarm rate and provide the best possible
outcome.
