Clustering wireless camera sensor networks based on overlapped region detection

According to increase of multimedia applications such as video surveillance and environment monitoring in Wireless Multimedia Sensor Networks (WMSN), energy conservation and prolonging network lifetime become increasingly important. Therefore, energy conservation and maximization of system lifetime are commonly recognized as a key challenge in the design and implementation of WMSNs. In these networks, nodes sense the area with high degree of correlation due to the overlapping Field of Views (FoVs) and causes wasting network energy. We present “overlapped region detection clustering” as a new clustering method to reduce redundant video streams and overcome waste of the energy. We apply scene cut technique to obtain overlapped region between images and then we apply our new clustering algorithm using scene cut detection results. We also propose a new camera activation algorithm to maximize gained information from correlated cameras and balance the communication load. Simulation results demonstrate that applying our clustering and camera activation algorithms to nodes, the network communication load and energy consumption are effectively reduced at their maximum by 75% and 66.6% respectively compared to correlation-based scheme.