On Eliminating Static Shadow False Alarms in Automatic Incident Detection Systems
Mohamed Shehata, M. Sohail Pervez, Tyson W. Burr, Jun Cai, Wael Maged Badawy, Ahmad Radmanesh · 2006
This paper presents an adaptive empirical algorithm which identifies static shadows within video sequences and produces static shadow maps that are used to improve the performance of video based automatic incident detection (AID) systems. The algorithm distinguishes between static shadows and other objects using background generation, motion detection, and three static shadow filters. The proposed algorithm has been tested on streams from 9 cameras to demonstrate its detection accuracy and robustness in varying lighting conditions