A multi-dimension information processing method for low-power forest surveillance systems

Abstract To identify early-stage forest fires using ground-based sensing systems, this paper studies the radiation characteristics of smoke particles and interference particles under natural light in different environments based on a rapid dimensional reduction algorithm of forest monitoring images and a multi-dimensional time series signal graph model. It has completed the forest fire classification and spread prediction method based on multi-sensor and image sequence signal abnormal event detection, which enables the forest fire detection system to achieve comprehensive and efficient collection of fire information at a lower cost and indirectly reduces sensor power consumption and bandwidth. The results can provide quantifiable solutions for forest fire detection and disaster situation assessment and are suitable for the comprehensive promotion of on-site sensing systems.