Flexible clustering in networks of smart cameras

Advances in sensors, networking and embedded computing have driven a paradigm shift in the control of camera networks. The processing of the image data has migrated from central servers to the camera network following fully distributed or clustered approach. In clustering, the data is analyzed within a group of cameras, and the abstracted data is then forwarded to other clusters. In this paper we present a flexible and scalable software tool that supports clustering in smart camera networks. This tool supports the dataflow-oriented processing within a group of cameras. A plug-in-mechanism facilitates flexibility and scalability; data processing can be migrated between the cluster node and the smart cameras. We demonstrate the feasibility of our tool in a multi-camera person tracking case study.