Detecting Uncommon Trajectories

An effective video surveillance system relies on detection of suspicious activities. In recent times, there has been an increasing focus on detecting anomalies in human behaviour using surveillance cameras as they provide a clue to preventing breaches in security. Human behaviour can be termed as suspicious when it is uncommon in occurrences and deviates from commonly understood behaviour within a particular context. This work aims to detect regions of interest in video sequences based on an understanding of uncommon behaviour. A commonality value is calculated to distinguish between common and uncommon occurrences. The proposed strategy is validated by classifying walking path of the people in a shopping mall corridor. CAVIAR database is used for this purpose. The results demonstrate the efficacy of the proposed approach in detecting deviant walking paths.