Comparative Study on Early Smoke Detection Based on Video of Forest Fire Front and Side Shooting

  • Wanjun Yu, Naimeng Cang

Abstract

Traditional smoke detection uses contact detection methods such as smoke, temperature, and light, which have limitations in space and scenes. Video-based fire detection and recognition is the current research hotspot, and fire monitoring and recognition based on multi-rotor aircraft is a new research direction. This paper uses the background   blur characteristics of smoke to compare the two surveillance videos. First, the background of the video is extracted by establishing a background model, and then the frame images of the video are decomposed and reconstructed by two-dimensional discrete wavelet transform to complete the recognition of smoke. The experimental result is that the average recognition rate of smoke from remote side shots is 86.83%. The average smoke recognition rate of the aircraft is 96.94%.

Published
2021-03-01
Section
Articles