Research On Fire Detection Method Based On Video Smoke Motion Detection

Fire has serious threats to human life,property,and natural ecological environment,It is of great significance to reduce various losses by detecting fires and giving early warning.With the popularity of intelligent monitoring equipment,Video-based fire detection technology has received extensive attention.Compared with the traditional fire detection methods,this method has the advantages of being free from environmental constraints,fast response,and wide detection range,and is suitable for fire monitoring in large space places such as shopping malls,factories,and forests.However,at the beginning of the fire,because the object is in a smoldering state,often accompanied by the production of smoke,detecting fires early by detecting smoke,therefor,video-based smoke detection technology has become a research hotspot.However,the current video smoke detection technology still has the problems of incomplete extraction of smoke regions and high rate of false alarms in smoke detection.Therefore,this article focuses on the above problems.The specific research work is as follows:(1)A method for extracting suspected smoke regions based on two steps image segmentation and motion object detection is proposed.First,using Otsu algorithm to perform two steps segmentation of the image to obtain a region of interest containing smoke;Then the visual background extraction algorithm is used to detect the motion of the smoke and the region search strategy is used to extract the suspected smoke region;Finally,using the image morphology processing method to obtain the suspected smoke region;The experimental results show that this method can extract more complete suspected smoke region than other methods.(2)Developed video smoke detection algorithm based on smoke multi-feature fusion.Aiming at the interference of pedestrians,cars and other objects in complex scenes,the motion features such as absolute smoke region growth,smoke diffusion,and the static features such as color,background blur,and texture that reflect the nature of smoke region extracted.Through the fusion of the extracted static and dynamic features,the SVM support vector machine was trained and the smoke detection classification model was obtained.The algorithm was tested with smoke and non-smoke video.The experimental results show that the algorithm is feasible.(3)A video smoke fire detection software was designed and an experimental platform was set up.Designed a fire video smoke detection system based on a research video smoke detection method,Modularize the algorithm in this article,based on MFC framework programming,software module functions.The software was tested by firing cigarette cake.The test results show that the software can detect the smoke on the video...