100 cameras each with one million pixel resolution operating at 30 frames per second would require 24Gbps of bandwidth without data compression. Even at 100:1 compression the bandwidth is still large relative to existing digital systems, and this bandwidth is continuous. We propose a technology that continuously digitizes and records data from a scaleable system with 1000's of cameras. This surveillance system uses novel digital cameras which greatly reduce bandwidth and data storage requirements while increasing visibility and lowering system cost. The proposed cameras will use complementary metal oxide semiconductor (CMOS) imager integrated circuits (ICs) designed both to capture and process images individually and to perform tracking and event triggering functions. The ICs will include efficient image comparison circuits to identify areas of interest in the surveillance field of view (FOV). Overall cost effectiveness derives from integrating new functions on the imager IC to provide a 'smart' camera and reduce infrastructure requirements. Phase I will analyze, specify, and design a CMOS imager IC meeting the system requirements. Phase II will manufacture the imager chips and demonstrate the technology in a portable system using a single computer and a single serial interface servicing up to 100 cameras.
