Smoke Detection Method Based On Smoke Density With Video

In recent years, smoke detection technology based on video is a very populardirection in the field of fire detection. Many researchers from home and abroad have madedeep research in this direction presenting different effective ways in fire safety alarm. Butmost of these methods are proposed on the basis of image blurring caused by smoke andsmoke diffusion. They can’t reflect the essence of smoke. So these methods may makewrong judgments. This article analyzes smoke density which is the one of the essence ofsmoke and proposes a new method to detect fire by smoke density with video to improveprecision.The article presents a thorough study of scattering and absorption caused by smoke andestablishes an optical model to analyze the image enhancement caused by scattering andimage attenuation caused by absorption. Then, the article deduces the relationship betweenthe current video frame, environmental parameters and the concentration of smoke andcomes up with the method to detect smoke density by video. After that, we verify themethod to detect smoke density by video images through comparing the results getting byvideo and the measured value by smoke meter in the different light intensity andbackground conditions.The article put the method which detects smoke by video into the detection of firesmoke. The article gets moving regions and updates the background through thecombination method of background difference and frame difference based on OTSU toupdate background and reduce processing region. After that, the binary image of movingregions is processed by median filtering and morphology method to remove the isolatednoise in the image and fill the moving regions. Then, the article removes noisy region bycircularity and structural similarity to extract suspicious area which provides an effectiveinput for analysis of smoke density. At last, we calculate the smoke density of suspiciousarea by the formation of haze image and study the variation of smoke density with time inorder to detect fire and give an alarm judging by the threshold of the variation of smokedensity and smoke density.The article test to verify practical applicability of the way based on smoke density indifferent environmental conditions. Then we analyze and conclude the experimental datato improve algorithm.