Optimized stereo reconstruction using 3D graphics hardware

In this paper we describe significant enhancements of our recently developed stereo reconstruction software exploiting features of commodity 3D graphics hardware. The aim of our research is the acceleration of dense reconstructions from stereo images. The implementation of our hardware based matching procedure avoids loss of accuracy due to the limited precision of calculations performed in 3D graphics processors. In this work we target on improving the time to obtain reconstructions for images with small resolution, and we covered other enhancements to increase the speed for highly detailed reconstructions. Our optimized implementation is able to generate up to 130 000 depth values per second on standard PC hardware. Additionally we extend the original epipolar reconstruction approach to estimate disparities between two images without known orientation. Therefore we can calculate e.g. the optical flow between successive images using 3D hardware.