The Research On Multi-target Tracking Algorithm Based On Video

With the development of the intelligent city and the public safety issues growing important,urban street installation of the camera also increased,the method of manual monitoring obviously has been unable to meet the development of urban security,intelligent video surveillance will become the mainstream of future development.The target tracking algorithm is the main content of intelligent video monitoring,which is used to locate the continuous position of the target in the video sequence,provides important information for the follow-up behavior recognition and abnormal behavior analysis.Target tracking can be divided into single target tracking and multi-target tracking based on the number of objects of interest Due to the limitation of single target tracking,multi-target tracking is becoming a hotspot..In this paper,the multi-target tracking problem in intelligent video surveillance is studied and discussed.The main contents and innovative achievements are as follows:1.The research on the moving target detection algorithm.The advantages and disadvantages of several kinds of mainstream detection algorithms are compared experimentally.The problem of low aging detection of DPM algorithm is studied and improved.Through the cascade detection model and convolution code optimization,the algorithm is improved in terms of time efficiency.2.The single target tracking method is discussed,and the experimental comparison and analysis of each mainstream algorithm are realized.3.The multi-target tracking method is studied,and the method based on energy function model is mainly studied.In order to improve the robustness of the tracking algorithm,the similarity model is added on the basis of the original energy function model,which can provide important cues for disambiguating it from the background and from other objects,so as to improve the accuracy of the tracking,and the experimental comparison shows the effectiveness of the algorithm.