The evolution of video surveillance: an overview

Abstract

Over the past 10 years, computer vision research has matured significantly. Although some of the core problems, such as object recognition and shape estimation are far from solved, many applications have made considerable progress. Video Surveillance is a thriving example of such an application. On the one hand, worldwide the number of cameras is expected to continue to grow exponentially and security budgets for governments, corporations and the private sector are increasing accordingly. On the other hand, technological advances in target detection, tracking, classification, and behavior analysis improve accuracy and reliability. Simple video surveillance systems that connect cameras via wireless video servers to Home PCs offer simple motion detection capabilities and are on sale at hardware and consumer electronics stores for under $300. The impact of these advances in video surveillance is pervasive. Progress is reported in technical and security publications, abilities are hyped and exaggerated by industry and media, benefits are glamorized and dangers dramatized in movies and politics. This exposure, in turn, enables the expansion of the vocabulary of video surveillance systems paving the way for more general automated video analysis.