Research On Target Handoff In Multi-Camera

Owing to limited vision, single view, the application of single camera exposes a number of shortcomings. It can not monitor huge scene, and has difficulties in solving the occlusion problem. The emergence of multi-camera, therefore, is inevitable, consequently, imperative researches on continuous tracking of the pedestrian in multi-camera, gait analysis, virtual reality etc, should be done. The application of multi-camera solves the problems that exist in single camera, at the same time, it raises new problems. Therefore, the research on target handoff in multi-camera has important theoretical value and prospect.In this thesis, the target handoff in multi-camera is the core of the study. The main work of this thesis is as follows: First, target detection and tracking problem under single camera are analyzed, which are the premise and basis of multi-camera research. Background subtraction method is used in targeting detection and the background update algorithm of right value which is joined to adapt to the changes of background scene. The combined algorithm of Mean Shift and Kalman Filter is used to track target. The algorithm has proved to be good real time and robust by large number of experiments. Second, some key algorithms and techniques about target handoff in multi-camera are studied in depth, recovering algorithms about camera field of view lines based on Harris and homograph is proposed in this thesis, which is thought to draw image registration. The various steps and related experiments results about field of view line are introduced, and experimental results show that our method is accurate and reliable. Finally, target handoff algorithm based on the field of view line is realized. Considering distance calculation algorithm proposed previously, we conducted a program that is a combination method of distance calculation and location information.Experiments show that the improved algorithm is simple and easy to implement, and can meet the requirements of real time, target handoff rate is 86% or more, to which show the algorithm can complete the goal of target handoff process accurately.