A Credibility Scoring Algorithm to match surveillance video target and UWB tag

This is the authors' abstract. We don't add key points for this paper.

Abstract Pedestrians positioning by surveillance video offers high accuracy and real-time performance in indoor scenes. Pedestrian positioning based on surveillance video has both challenges and promising application prospects. However, the practical application of existing surveillance video is hindered by the lack of tag information and incomplete video coverage. Although the positioning methods with terminal devices provide the tag information, simultaneous signal transmission from multiple tags to the base station will lead to collisions and mutual interference, like UWB positioning. The positioning method that combining surveillance video and terminal devices is expected to solve the problem. This paper introduces a credibility scoring algorithm to match surveillance video target and terminal tag. The algorithm considers four factors, including distance between the obtained positioning points, weighted average speed, direction of motion, number of obtained positioning points from both the video and Ultra-Wideband (UWB) tag. The algorithm matches the surveillance video target and UWB tag according to the highest sore. The experimental results show that the proposed credibility scoring algorithm achieved accurate matching results in both unob-structed scenes (scores for the same tag are generally twice as high as scores for different tag) and obstructed scenes (scores for the same tag are 13% higher than scores for different tag). The approach improves the usability of pedestrian positioning by surveillance video in indoor scenes and avoid the problem of tag mis-alignment, and will further provide the technical support in pedestrian positioning and management based on surveillance video.