Research And Implementation Of Human Target Tracking System For Intelligent Monitoring

With the development of the times,people’s consciousness has been improved,and safety awareness has also become more and more popular.As an extremely important security measure,video surveillance is an important guarantee for people’s safety.At present,most video surveillance systems in life rely mainly on dedicated staff to view the situation in the video in real time,which has the characteristics of low efficiency,untimely processing,and large blind spots.Therefore,the development of an intelligent video surveillance system that automatically monitors the video in surveillance cameras is of great significance for maintaining social stability and ensuring people’s safety.Human target tracking technology is the key to realizing intelligent video surveillance system,and its result will directly affect the monitoring effect of intelligent video surveillance system.The existing human target tracking algorithm is only suitable for video tracking in simple scenes,but because it ignores the mining and consideration of human features in surveillance scenes,the tracking effect on complex surveillance videos is not very ideal.Therefore,this topic introduces human detection into the target tracking framework.First,the human detection in the monitoring scene is studied,and then the human target tracking algorithm based on human detection is studied.The specific research contents of this article are as follows:First of all,this paper analyzes the reasons for the difficulty in human detection in surveillance video.The real surveillance video is obtained from the Public Security Bureau to produce a human video detection data set.The VGG16 classification network is used to realize the parameter migration of the detection network and optimize the adaptation.For non-maximum suppression of human detection,an optimization algorithm for candidate region generation based on the three-frame difference method is proposed to implement a human detection method for surveillance video.This method effectively improves the accuracy of human detection in complex surveillance scenarios.Then,in view of the current human target tracking algorithm ignores the mining and consideration of human features in real surveillance scenarios,this paper introduces the human detection method for surveillance video proposed above into the target tracking framework,and proposes a robust for complex video surveillance Human target tracking method.Based on the SiamMask network,this paper optimizes the target model of the initial frame and the selection method of the search area,and proposes a method for calculating the weight of the target tracking candidate region based on human detection.In addition,a fusion of human detection results and target continuity between adjacent frames The update method of the target model.This method effectively improves the accuracy and robustness of human target tracking.Based on the above research content,this paper designs and implements a human target tracking system.Through a large number of experiments,the effectiveness of the proposed method is analyzed and verified from the two perspectives of subjective performance evaluation and objective data statistics.The experimental results show that the human target tracking algorithm proposed in this paper is affected by many factors such as illumination and distance.The complex monitoring video can still maintain a good tracking effect,which improves the performance of the existing human target tracking algorithm in the monitoring scene.