Editorial introduction to the special issue on "Image Understanding for Real-World Distributed Video Networks" - Computer Vision and Image Understanding Journal

Cities around the world are increasingly relying upon cameras to provide visual coverage of extended spaces. Surveillance cameras are an essential component of the overall crime prevention strategy. Law enforcement agencies now heavily use imagery collected by surveillance cameras in solving crimes. Cameras installed on roadways are used to collect traffic data with the aim to better manage traffic, relieve congestion, respond to accidents, etc. Surveillance cameras are also used in high-stake public places, such as airport, train stations, metro stations and bus terminals. In recent years, cameras have also been used to record and prevent incidents of police brutality. The current security climate has necessitated the need of increased monitoring and surveillance of citizenry. In short, as a society, we are increasingly reliant on cameras. Advances in camera hardware and communication technologies have enabled us to set up camera or video networks providing surveillance and monitoring capabilities over large areas. In the past, video collected from these cameras was monitored by human operators trained to view these videos and respond appropriately to any incidents visible in these videos. In many cases, the video was simply recorded somewhere and it was accessed for forensic purposes only after an incident has already occurred. The capacity to ''observe and analyze'' data collected by all these cameras simply did not exist. Automatic analysis of video collected from these cameras is an important capability and consequently there is significant interest in the computer vision community to develop theory and techniques to automatically analyze and understand data collected from video networks. It is helpful to remind ourselves that video networks are fundamentally different from multi-camera systems of yore. Multi-camera systems often assume that the data collected by each camera is collected at a central location where it is available to the video analysis algorithms. This assumption does not hold for video networks. Simply because the sheer scale of video networks necessitates the use of distributed techniques for video analysis. These techniques do not assume that video from cameras is gathered at a central location and that all of it is available to the video analysis algorithms. This special issue deals with computer vision techniques that are especially suited for automatic video analysis in distributed video networks. Distributed video networks is an exciting and emerging multidisciplinary field that draws upon areas as diverse as, sensor networks, distributed systems, embedded devices, computer vision and pattern recognition, networking, …