Rapidly Deployable Distributed Video Surveillance System for Resource Constrained Applications

We report on a the design of a video surveillance system designed for rapid deployment on devices with limited computation and communication capability. The system employs a new video encoding algorithm specifically designed for surveillance of moving objects in a slowly changing environment. The prototype system is deployed on a mixed Linux network, with video streams originating from desktop, laptop, and Intel PXA25S-based embedded devices. Low computational demand and high compression rates are achieved by segmenting foreground objects from the background and streaming only meta-information and compressed imagery representing the foreground. Rapid deployment and seamless monitoring are provided through a web-based interface that manages multiple geographic regions and the cameras they contain, camera addresses, motion detection alerts, and video streams.