| China has a vast territory,complicated geographical and climatic conditions,and frequent forest fires.The real-time monitoring of forest fires has become a focus of attention,and early smoke detection of forest fires has more research value than flame detection.Traditional smoke detection uses contact detection methods such as smoke,temperature,and light,which have limitations in space and scene.Video-based forest fire detection and recognition is the current research hotspot.With the rapid development and popularization of UAVs,fire monitoring using multi-rotor aircraft is called a new research direction.The simulation experiment video in this article is through UAV taking pictures.This paper not only uses the background blur characteristics of smoke for comparative research,but also uses the HSV color model to extract the color characteristics of smoke.For the former,this article first establishes a background model to extract the video background;according to the high-frequency attenuation characteristics of smoke in the frequency domain,the image is wavelet decomposed,then highfrequency information is fused,and the fused image is compared with the high-frequency information of the background image of the corresponding video frame If the high-frequency information of the local area is reduced,it is determined to be a suspected smoke area,and the largest connected area of suspected smoke is marked;the simulation experiment results show that the average recognition rate of smoke from remote side shots is 86.83%,based on the aerial photography of multi-rotor aircraft The average smoke recognition rate is 96.94%;therefore,this method can accurately detect the smoke in the early forest fire smoke video frame image,with a low false detection rate,and can reduce the influence of some interference factors on the detection accuracy of the early forest fire smoke video image.Aiming at the color characteristics of the smoke,this paper first converts the original smoke video frame image to the HSV color space,segmentation and combination segmentation are performed on each component,and the corresponding smoke area is determined according to the characteristics of the smoke in different components. |