Smoke Detection and Recognition Based on Flame Feature Extraction

The sensor-based fire monitoring scheme requires the installation of temperature and smoke detectors near the monitoring target. Meanwhile, the pictorial fire detection technology is capable of capturing the location and development degree of the fire instantly, enabling rapid detection. Aiming at the issue that the traditional fire monitoring system is prone to failure in large-space indoor and open outdoor environments, a digital image fire monitoring algorithm based on the characteristics of fire flames and smoke is proposed. In the early stage of a fire, smoke is frequently accompanied by its generation and can spread over a large area in the space and is not easily blocked. Therefore, the detection of smoke can realize the early warning of fires and conduct fire rescue work in a targeted manner. In this paper, the background subtraction method is initially employed to extract the moving object, and then the required flame model features are extracted through image preprocessing, and the fire image is identified through multi-feature fusion. The experiment indicates that the flame identification method proposed in this paper can accurately identify the flame under various circumstances and provides a favorable judgment basis for fire alarms.