Region And Active Shape Trackor For People Tracking System

This paper mainly is to select region and active shape tracking algorithms to achieve robust tracking of people in an indoor environment fro our bank. The tracking system contains two co-operating trackers: i) an Active Shape Tracker using a PCA -generated model of pedestrian outline shapes, ii) a Region Tracker, featuring region splitting and merging for multiple hypothesis matching. Data from the two tracker are fused together to select the best tracking hypotheses.A good realtime tracking system must possess 4 advantages: i)high speed and low computative cost, the system should process realtime video, and can be widely used, ii) good robustness: a good level of tracking robustness is achieved, people are always labeled the same one as long as he stays in the video. iii)good generality: the system can be used in different environments by adjust the parameter system.iv)to fulfil the demand of different stages or intension, the system can be works either standalone or as a subsystem of the integrated visual surveillance system.Consider the 4 factors above, a tracking system which was called Reading People Tracker was taken as the prototype of the system.This paper main modified the rigion tracker and the active shape tracker.Temperal integration static objects to the background, multiple hypothesis matching and multiple trackors inaction were applied to deal with the cases of occlusion and low contrast, lost and regained tracks.So a new experimential tracking system was build for the indoor enviroment of the bank. Experiment showed the new system work quiet well.The system is validated using sequences from surveillance cameras in a bank station. It is demonstrated that robust realtime tracking of people can be achieved with the new tracking system using standard PC hardware. And the system can be integrated with the existing surveillance system of the bank.