Analysis of shape features of flame and interference image in video fire detection

Abstract

In this paper, the shape features of different material of flame and interference image in video fire detection were analyzed and compared. The area, perimeter, rotundity, solidity, extent, sharp corners, similarity and centroid displacement are selected as the candidate features. After analyzing and excluding area and perimeter as these candidate features, weak classifiers of six other shape features are established, and then calculate weights of each weak classifier using AdaBoost algorithm, and then sort according to the weights of weak classifiers, and then proposed the preference of choosing shape features. In this paper, shape features selection preference is introduced, which can be used as a reference for video fire detection choosing shape features.