Optimized visualization of stereo images on an OMAP platform with integrated parallax barrier auto-stereoscopic display

In this paper, we describe a system for optimized visualization of stereo images on a mobile platform. The system utilizes a front camera, and face and eye tracking to find the position of the observer's eyes. Depending on this position, the left and right views targeting the corresponding eyes are maintained properly based on measured optical characteristics of the used parallax-barrier 3D display.An efficient implementation on the OMAP 3430 platform is targeted by splitting the processes of face and eye detection between the ARM and DSP cores. The final system allows for dynamic switching the display between 2D and 3D mode and swapping the left and right views so to avoid high cross-talk between view channels and reverse stereo effect.