A survey on performance evaluation of moving object detection and tracking in video surveillance system

초록

Performance evaluation has become an indispensable part in video surveillance systems, as researchers devote themselves to attain reliable and robust video surveillance systems with sufficient accuracy for practical applications. Unfortunately, the lack of wide-ranging metrics which are not inducing any bias towards evaluation results makes it hard to do evaluations on the surveillance system. Therefore, this paper reviews a set of metrics and algorithms to allow for objective comparison using frame-based, object-based, region-based and feature-based evaluation paradigms. The reviewed metrics are applicable for a wide range of detection and tracking system and allow for objective comparison of main characters