The Detecting And Tracking Methods Of Moving Target In Coal Mine Tunnel

The coal mine video operations have many problems such as complex environment,large noise,uneven illumination,shades,and the high rate of false detection.The existence of dangerous areas in coal mine tunnel requires a strict security management system.For coal mine tunnel safety production,moving target detection and tracking based on video surveillance is of great significance.In this paper,we introduced a new way to detect and track personnel targets under coal mine from videos and images.Because of the special underground environmen,t,the light of coal mine tunnel is uneven and the video lacks color information.In order to improve the video's quality and also remove the effect of dust,this paper introduced a method which is based on the combination of wavelet transform method and Dark Channel Prior method.The new method begins with the treatment of image using wavelet transform method,then we make a fogging treatment using Dark Channel Prior method,next we do a denoising process on high frequency coefficient using Semi soft threshold filtering method,finally,we combine the coefficients to restructure wavelet analysis.This method can effectively remove the image noise while enhancing the details of the image;at the same time,it solves the problem of dense fog,uneven illumination and uneven image contrast.For the problems of error detection and missed detection that come from blurring target and boundary,this paper introduced an improved Gauss background modeling method.It divides the video frame into blocks,replacing the current pixel value with the mean value of each block.Then adaptively assigns different number of Gauss distribution and update rate to each pixel block,and combines the mixed Gauss model and the three frame difference to detect the moving target in the video.In this method,a new update strategy is used to quickly model the background,and the pixel point stability is used to adjust the update speed of the pixels,thus reducing the computation of the algorithm,improving the speed of the algorithm,and meeting the requirement of real-time.Besides,a tracking algorithm based on spatio-temporal context is applied to track moving targets in the surveillance video.The simulation results show that the combination of wavelet transform method and Dark Channel Prior method can improve the quality of original videos and images.From the perspective of information entropy and average gradient,the improved algorithm is superior to the traditional enhancement algorithm.The improved Gauss background modeling method can realize the detection of humans under coal mine while using a lower detection error rate.That method has a certain theoretical significance for the low illumination coal mine image recognition and personnel positioning,which provides a pratical basis for the safety coal mine production.