Stereoscopic 3-D acquisition, processing, and display for telerobotic applications

Abstract

We present various solutions developed through our research to problems arising in the stereoscopic 3-D visualisation process for telerobotics applications. We show that real-time of processing video imagery is required to rectify geometric distortion that can negatively impact the quality of depth perception; that rectification can be achieved efficiently using both specialised hardware and commodity hardware such as graphic card GPUs; that the solution to the computationally intensive problem of real-time computational depth estimation can be speeded up using commodity graphics card MPEG encoders; and that the problem itself can be simplified through a novel scene illumination and image acquisition strategy. Finally, we show how a display device incorporating an adaptive optics element uses computed depth to display the 3-D scene with appropriate optical distance - thus avoiding a well-known cause of visual discomfort in stereoscopic 3-D visualisation. © 2007 Springer-Verlag Berlin Heidelberg.