Dynamic 3D Sensors: Data Characterization and Post- Processing

After a brief introduction about Time–of–Flight range cameras and 3D sensor of Microsoft Kinect characteristics, a deep analysis on statistical distribution of data retrieved from these sensors, is performed. A set of algorithms and procedures are designed and implemented to improve the general quality of the depth–maps acquired, on the basis of the problems highlighted. They are computed in particular denoising and upscaling operations, through the use of an innovative and smart smoothing filter, the trilateral filter. The main attention is focused towards Kinect sensor, but the procedure can be adapted to other setting of utilizations. In the end they are presented experimental results, applications and further improvements