Assessment of video analytics for exterior intrusion detection applications

Abstract

When combined with a compatible assessment system, select video analytic devices can provide intrusion detection and target-tracking capabilities. With an attractive relative low cost, demand for video analytics has focused on high probability of detection and low nuisance alarm rate that has prioritized investment toward performance improvements. Modern algorithms offer sophisticated video-based intrusion detection and tracking capabilities that were unavailable a decade earlier. The modern video analytics have been shown to have a significant reduction of previously observed high nuisance alarm rates with improved detection capabilities. This paper includes results of a recent device evaluation performed at Sandia National Laboratories for a representative exterior intrusion detection application. For this limited scope effort, the device was evaluated for probability of detection and monitored for nuisance and false alarms over a six-month period from September 2010 to April 2011 (nuisance alarm cause is detailed). This paper also includes general and specific implementation considerations for video analytics. The author asserts that this work holds potential to improve implementation strategies for video-based detection sensors for several types of exterior perimeter and extended detection-sensing applications, making the case for interoperation and compliance standardization between video and sensor technology vendors for simplified integration of video analytic algorithms into security platforms.