Fire detection in different color models

Abstract

Detecting forest fire is a highly active research area in the field of pattern recognition and computer vision as a number of existing methods available in the literature. The purpose of the proposed study is to select the most suitable color space, features and classifiers for the fire classification. Our approach begins by finding the likelihood of every pixel value. The fire is then defined by multiplying pixel channel value's likelihood. By using a simple thresholding schema, the fire pixel classification process is performed. Our experimental study demonstrates which color space, features and classifiers are most suitable for fire detection.