Smart camera networks are real-time distributed embedded systems that perform computer vision using multiple cameras. They have emerged thanks to the simultaneous advances in four key disciplines: computer vision, image sensors, embedded computing, and sensor networks.

“We aim at advancing this field of research by applying novel networking concepts as well as by developing various prototypes,” Bernhard Rinner explains. Analyzing the captured data within the network in real-time is important to avoid transferring large volume of video data over the network. The strong resource limitations are challenging and require efficient algorithms and network management. The research team has developed various camera platforms and has deployed several camera networks both in indoor and outdoor environments. Prototype installations include traffic monitoring, environmental monitoring, and surveillance.

In-network processing is concerned about the management of the available resources in the camera network. Examples of such resource management include clustering, i.e., selecting a group of cameras which jointly work a specific task, handover, i.e., transfering a specific task from one camera to another, and calibration, i.e., estimating the spatial relationship among the individual cameras.

In her recent work, Jennifer Simonjan developed a decentralized and resource-aware algorithm for estimating the poses of all camera nodes without any user interaction. “Self-calibration is achieved in two steps,” she explains. “First, overlapping camera pairs estimate relative positions and orientations by exchanging locally measured distances and angles to detected objects. Second, calibration information of overlapping cameras is spread throughout the network such that poses of non-overlapping cameras can also be estimated.”

Bernhard Rinner and his team have been involved in smart camera research for more than a decade. The following articles document some highlights of that period: