Large-Scale Intrusion Detection with Low-Cost Multi-Camera Wireless Image Sensors

Abstract : We present a low-cost multi-camera wireless image sensor platform based on commercial off-the-shelf components implementing an intrusion detection mechanism for large-scale surveillance applications. We first show the performance of the image sensor then present how a criticality-based scheduling approach can be adapted and integrated into the image sensor platform taking into account the real hardware constraints and capabilities. We then demonstrate by simulating largescale configurations that although hardware constraints limit the maximum intrusion detection rate per node, the usage of a low-cost multi-camera system provides much higher detection quality than denser single-camera system working at a higher image capture rate.