Research On Key Techniques Of Multi Moving Target Tracking Based On Vision

This project predicts the moving direction and the movement state of the target in a short time according to the different effective features and movement tendency of the target,based on the detection of multi-moving targets.Providing the data source for the next target analysis.Multi-target tracking technology integrates many disciplines such as computer vision,image processing and pattern recognition.It has a high practical status in many fields such as military,intelligent monitoring and medical treatment,and can be regarded as a very important and challenge issue.It has a broad prospects for development.On the basis of deeply studying on the theory and application field of multi-moving target tracking,this paper focuses on the tracking technology of multi-moving target.The article mainly studies the key of technologies of multi-sport tracking based on multi-sport target recognition and multi-sport tracking.Researching on the multiple moving target detection technology.First of all,the paper analyzes the principle,applicable scenes and so on of all kinds of classical algorithms in multi-sport detection.Then,studying the methods of background modeling,background comparison,single Gaussian and mixed Gaussian background modeling and so no,comparing the experimental verification and analysis.The background modeling method of ViBe is emphatically studied and the simulation analysis is carried out.The existing main problems are improved,the improved ViBe background modeling and morphological processing are combined.The improved algorithm is applied to the detection process to achieve multi-moving target detection.Multi-moving target tracking algorithm research,first introduced the basic theory of multi-moving target tracking and some of the current classical algorithms in this field.Then,the key tracking algorithms,such as Kalman filter estimation,Meanshift algorithm and space scale change model,are researched and analyzed emphatically,and the experimental verification results are analyzed and compared.This paper improves the algorithm of Kalman filter forecasting unit,focuses on the change of size and crossover of Meanshift algorithm in tracking process,improves the algorithm by using spatial scale change model and Kalman filter prediction model,and verifies the accuracy of the algorithm by experiments sex,to achieve a multi-sport tracking.