Detection Of Human Targets In Coal Mine Based On Video

Coal mine safety production has always been a key topic of social concern.At present,most coal m,ines have installed video surveillance systems to avoid accidents.The existing coal mine monitoring system mainly analyzes the video manually,which makes it impossible to make timely judgments on the accident.Although some mines have installed i.ntelligent monitoring,there are problems of low monitor,in.g efficiency and high false positive rate.Therefore,researching the personnel detection algorithm suitable for underground coal mines plays an important role in ensuring the safety of miners.The underground light in coal mines is uneven and is interfered by dust such as coal ash,resulting in poor quality of monitoring video and low contrast,which affects the subsequent detection of humanoid targets.In orcder to improve this problem,this paper proposes an improved UnSharp Masking image enhancement algorithm.The algorithm firstly uses the bilateral filtering to decompose the source image to obtain the high-frequency image.Secondly,the high-frequency image is amplified by the adaptive gain function.Finally,the amplified high-frequency image is fused on the basis of the source image to achieve the enhancement effect.Experiments show that the proposed algorithm improves the image contrast,completely preserves the details of the source image,and better highlights the humanoid target.Aiming at the problem that the underground mine lamp is irradiated into the monitoring area and mistakenly believes that the personnel intrusion will produce a warning,this paper proposes a downhole humanoid target detection algorithm based on shape features.The algorithm first enhances the video image;secondly,the Gaussian Mixture Model method is improved with,the improved three-frame difference method to avoid the phenomenon of"cavitation" in the foreground target;finally,the interference of the miner's lamp is eliminated by judging the aspect ratio of the foreground contour.Experiments show that the proposed algorithm effectively removes the miner's interference and improves the detection rate of humanoid targets.In order to solve the problem of low detection rate of humanoid target detection algorithm in complex coal mine environment,this paper adopts a downhole humanoid target detection algorithm based on HOG and LBP feature fusion.The algorithm first extracts the HOG feature and LBP feature of the sample,and reduces the dimension of the HOG feature.Secondly,the two features are serially merged and input into the SVM to train the classifier model.Finally,the classifier parameters are optimized.Experiments show that the proposed algorithm improves the detection rate of humanoid targets in complex coal mine environments and improves the general applicability of the algorithm.