Video Based Target Trajectory Detection In Contaminated Remediation Sites

Due to the complex environment in the contaminated remediation site,the concentration of toxic gas in some areas is too high,which will cause harm to the health of interdependent humanity.Hence,how to efficaciously control the interdependent humanity,correctly obtain the movement track of the interdependent humanity,impede them from entering the dangerous area,and assure the humanity safety of the interdependent humanity,is the problem that must be faced with in the environmental remediation operation.In this paper,the travel trajectory of video target in the intellective surveying system of contaminated remediation site is studied,which mainly includes three stages: real-time detection,stable tailing and ballistic extraction.The chief job of this essay are shown below:(1)Targeted at the matters of fancy and changeable light in the place of contaminated remediation,the characteristics of five classical target detection algorithms,Gradient Flow,Salience Aware Geodesic,Frequency Tuned,Geodesic Salience and Manifold Ranking,are analyzed.Finally,a Video Salience Detection based on Prior Fusion model(VSDPF)is proposed to detect motion targets.The model uses a priori fusion technique,that can correctly suppress the background area,thus improving the integrity of foreground salient target gauging,and has high productivity,which contents the demands of actual-time system.(2)Targeted at the matter of fast size varies and low tailing productivity of moving targets in contaminated remediation environment,five classical tracking algorithms are discussed,including Multiple Instance Learning,Tracking Learning Detection,Median Flow,Kernel Correlation Filter and Boosting.Finally,fast Discriminative Scale Space Tracker algorithm(f DSST)is used to track the target in real time.Based on the scale adaptive visual tracking method,the algorithm uses feature reduction and interpolation techniques to speed up the algorithm greatly.After acceleration,the area of search box is increased and the accuracy is also improved.(3)Targeted at the move object,the center collection way is accustomed to gain the movement track of the construction personnel in the video sequence.This method can be applied in the intelligent monitoring system of contaminated remediation site.Experimental results show that the average absolute error(MAE)of the proposed VSDPF model in ViSal data set is about 0.096,and the MAE value in Segtrack V2 data set is less than 0.117,which can accurately detect video motion targets;the FPS(Frames Per Second)value of the f DSST model can reach 35.868 frames / s,which has high tracking efficiency for moving objects and increases the recognition rate of trajectory extraction.The algorithm has been applied in the intelligent monitoring system of contaminated remediation site,which meets the requirements of the system.