Research On Pedestrian Tracking Methods Using Wi-Fi And Camera

The routing equipment that provides free Wi-Fi Internet services and security camera that provides are now distributed in public areas.Video positioning technology has the advantages of accurate positioning and high real-time performance in lowdensity scenes.However,due to the limited coverage area of a single camera and the difficulty of processing across multiple cameras,video-based pedestrian tracking technology is difficult to deal with cross-domain tracking of pedestrians reidentification(Re-ID)problem.At the same time,Wi-Fi-based mobile device positioning technology has also received widespread attention.Although the positioning accuracy is limited,the Wi-Fi AP(Access Point)has a greater coverage than the camera,and the mobile device has a unique MAC address,with the result that the wireless positioning technology based on Wi-Fi generally does not need to tackle the Re-ID problem.Therefore,by integrating the two different positioning technologies,i.e.Wi-Fi and camera,it is not only helpful to solve the Re-ID problem faced by video positioning,but also to improve the accuracy of Wi-Fi positioning,thereby achieving more accurate pedestrian tracking.This paper proposes two trajectory matching algorithms to solve the Re-ID problem,designs and implements a set of video and Wi-Fi fusion positioning system for pedestrian tracking.First of all,by analyzing the characteristics of wireless positioning technology and video positioning technology,we can understand the similarities and differences between the positioning process and results.Secondly,two kinds of trajectory matching algorithms are proposed to match the localization results:(1)the data matching algorithm based on weighting motion states,which analyzes the motion state(speed,direction,displacement and other factors)between the objects generating the trajectories,corresponding weights are defined to measure the similarity between the trajectories;(2)the data matching algorithm based on the conditional probability model,which establishes the probability model describing the similarity between two trajectories.Thirdly,by analyzing the actual scene requirements,a video and Wi-Fi integrated positioning system is designed and implemented.Finally,experiments are carried out in the actual environment to verify the effectiveness of the trajectory matching algorithms and the usability of the fusion positioning system.In summary,two trajectory matching algorithms proposed in this paper can effectively match Wi-Fi positioning results with video positioning results.Moreover,the design and implementation of the video and Wi-Fi fusion positioning system can provide a reliable reference scheme for pedestrian tracking across camera views,which has a certain practical significance.