Research On Moving Target Detection And Tracking Algorithm In Intelligent Video Monitoring

In recent years,intelligent video surveillance plays an important role in both industrial production and security field.Moving object detection and tracking are the important parts of the intelligent video monitoring system.The accuracy,stability and real-time of the algorithms will directly affect the performance of the system.In this paper,ViBe algorithm and KCF algorithm are studied and improved,and the integration of detection and tracking is realized.The specific research contents are as follows:(1)Several mainstream moving target detection and tracking algorithms are studied.Their advantages and disadvantages are analyzed by experiments.Finally,ViBe algorithm and KCF algorithm are chosen as the main research objects due to their advantages.(2)For the ghost and the interference of dynamic background,an improved ViBe algorithm is proposed.For the problem of ghost,detect the saliency of the pre-M-frame.Find the moving area according to the saliency diagram and synthetic relatively real background to model.For the problem of poor disturbance resistance,a method of calculating background complexity is proposed,and the threshold is set according to the complexity adaptively.Finally,fill the inner contour of extracted object to make the target more complete.Experimental results show that the improved algorithm achieves better results in accuracy and robustness.(3)For the problems of shadow,a kind of method based on multi-feature adaptive background model is presented.Firstly,establish the multi-feature background model to improve the representational ability of the model by fusing chromaticity feature and texture feature.Then,adjust the model adaptively according to the background complexity and extract the moving object.Finally,update the background model.The test results show that the improved method can eliminate the influence of shadow well.(4)For the problems of scale fixation and occlusion,an anti-occlusion and scale adaptive improved KCF algorithm is proposed.For the problem of occlusion,the occlusion discrimination mechanism is introduced.When the target is in occlusion,stop updating the model to obtain a more accurate position.For the problem of fixed scale,samples of different scales are collected to calculate the response value.The scale corresponding to the maximum response value is the optimal scale.Experimental results show that the improved algorithm can accurately track the target in the sene of scale transformation and occlusion.(5)In order to meet the actual needs,this paper realizes the integration of detection and tracking.The results of the detection algorithm are input into the tracking algorithm to realize automatic tracking.