Depth stream compression for enhanced real time fall detection by multiple sensors

A fall detection system using depth data acquired from a single Kinect sensor could be expanded by addition of more devices. The idea consists in equipping each sensor with an embedded board, in order to stream the collected depth data over a network, towards a central system, able to retrieve the data and process them. A preliminary evaluation of a low-complexity compression algorithm for depth streams is performed, and some encouraging results are presented.