A Research Of Moving Target Tracking And Handover With Multi-cameras

The traditional monitoring technique using single camera cannot meet people's needs due to the narrow field of view and difficult information fusion.In recent years,multi-camera-based intelligent video monitoring systems have attracted widespread attention from the academic community and industry.Camera deployment modes and the states of the target in time and space are diverse.As a result,how to extract,detect,track,and handover moving targets with multiple cameras is the core and key issue in multi-camera monitoring systems.To solve the addressed problems,this paper designs and implements a multi-camera-based moving target detection and active tracking system,realizing real-time monitoring and target tracking requirements in multi-camera background.Using a moving target detection algorithm,the moving target is separated from the background.Afterwards,the moving target is locked and continuous tracking of the moving target is achieved.After the target enters the second camera's field of view from the current camera,the consistency of the target is determined based on the overlap of the two camera's field of view.After the two cameras complete the correct and stable handover of the target,the target continues to be tracked in the second camera.The main work and contributions of this paper are summarized as follows:1.This paper studies the target handover technology based on the fields of vision.Compared with the traditional method,this paper utilizes an algorithm based on YCbCr color space histogram matching and the visual field dividing line,which can effectively solve the misjudgment problem that the targets are too close to be tracked correctly.If the overlapped field of vision is too small,the traditional method will also have the error matching.This paper proposes an adaptive target handover algorithm based on visual field boundary and feature matching which will solve the problem.The field of vision between multiple cameras is generated by the method of SIFT and homography.2.This paper studies the moving target extraction and detection algorithm.Using this algorithm,the moving object is extracted and detected from the background image in the video sequence.With the separation of the foreground image and the background image,information such as the position and motion status of the moving object is obtained.3.This paper studies moving target detection and tracking technology.After analyzing the commonly used classical algorithm,we select the suitable moving object detection algorithm for this paper,which will provide a good foundation for subsequent target tracking and handover.4.This paper sets up an experimental platform for verification and completes the dual-field motion target tracking and handover experiments.Using C++ language on VS2010 and opencv 2.4.9 platform,the system realizes several modules,such as target detection module,tracking module and target handover module.