The Research And Realization Of Mine Personnel Target Detection And Tracking Algorithm

The coal mine has been equipped with a relatively complete video surveillance system,but it relies on manual inspection,and the omission of inspection is likely to leave safety hazards.Using intelligent target detection and tracking technology to process mine videos can detect potential safety hazards and issue warnings in time,which has important practical significance for mine safety.The specific work of this paper are as follows:1.Use the background difference method to detect the target,carry out the mixed Gaussian background modeling,and dynamically update the background.Miners wearing miner’s lamps and flashlights will form dynamic light spots,which are misjudged as targets during detection.A dual-threshold background difference method is proposed to remove light spots;the shadow parts of miners are judged and deleted by combining the pixel threshold and a correlation coefficient method.The experiment proves the feasibility of the algorithm.2.Aiming at the deficiencies of the Camshift algorithm in tracking,the dual-threshold background difference method and Kalman filter are combined to improve the Camshift algorithm,and Gaussian weighting is introduced.When it is detected that the target has occlusion or interference,the Kalman prediction value is used as the observation value,and the Camshift tracking value is used as the observation value if it does not appear,and the target template is updated.The improved Camshift algorithm can effectively overcome the problems of occlusion and dark background interference.3.An improved target tracking algorithm based on particle filter and adaptive multi-feature fusion is proposed.The algorithm combines color and texture features,and can adaptively adjust the weight coefficients of features.Improve the traditional particle algorithm,first optimize the particle weight,increase the weight of low-weight particles;when resampling,limit the number of particles after re-sampling of large-weight particles,and appropriately increase the number of re-sampled particles of small-weight particles to increase particle diversity.This tracking algorithm can effectively overcome the light changes in the mine,the dark background and the color interference and occlusion of the miner’s clothes,and it has good robustness.