Indoor occupancy logger with compressive vision sensing

An occupancy logger that determines user presence and estimated locations in an indoor space may be used for occupancy analytics, space management and other building services. We consider use of a vision sensor along with compressive sensing techniques for occupancy logging. In the presented solution, a sparse representation of captured images is obtained at the sensing device. The sparse representation is sent to a cloud computing unit (CCU). At the CCU, occupancy changes are determined using a change detector. If occupancy is determined, the image is reconstructed using compressive recovery techniques and occupancy is localized and subsequently tracked.We evaluate the performance of this technique with a simple image differencing method under different illuminance conditions.