Making Time-of-Flight range imaging cameras applicable for human-machine-interaction by depth correction

For various applications, such as object recognition or tracking and especially when the object is partly occluded or articulated as in human-machine-interaction scenarios, 3D information is crucial for the robustness of the application. Recently sensor had been developed that acquire distance information based Time-of-Flight (ToF) principle and for which distance errors based on different causes can be observed. There are several technologies present for range imaging such as the Photo Mixer Device (PMD) working on modulated, incoherent infrared light or by using the depth imprint of an emitted light pulse by fast shuttering. This article presents a distance calibration approach for distance sensing which handles objects with different Lambertian reflectance properties because different albedos strongly affect the distance offsets. It combines the depth noise estimate and the intensity for a reflectivity estimate. The calibration results of the proposed method excel the results of other known methods due to the fact that they mostly do not incorporate the major effect of object reflectivity for the depth correction. In particular with objects with unknown reflectance properties a significant reduction of the error is achieved.