Nighttime fire smoke detection system based on machine vision

A real-time machine vision-based nighttime fire smoke detection method that can be incorporated into a surveillance system for early fire alert is proposed in this paper. Nowadays, video surveillance-based early fire smoke detection is crucial to the prevention of large fires and the protection of life and goods. However, many of the known video smoke detection methods require that minimum illumination be provided for the cameras to recognize the existence of fire smoke in a scene. To overcome the nighttime limitations of video smoke detection methods, a laser light is projected into the monitored field of view, and the returning projected light section image is analyzed in order to fire smoke. If the fire smoke appears within the monitoring zone created from the diffusing or scattering projected light path, the camera sensor receives a corresponding signal. The successive processing steps of proposed real-time algorithm are using the spectral, diffusing, and scattering characteristics of the fire smoke regions in the image sequences to register the possible smoke position in a video. Characterization of smoke is carried out by calculating accumulative deviation from the extracted feature vectors, and a classification method using a fuzzy reasoning system is applied to assign a score to the potential fire smoke candidate. Experimental results in a variety of nighttime conditions demonstrate that the proposed fire smoke detection method can successfully and reliably detect fire smoke.