作者: S. Forstmann , Y. Kanou , Jun Ohya , S. Thuering , A. Schmitt
关键词: Artificial intelligence 、 Computational science 、 Computer vision 、 Machine vision 、 Matching (graph theory) 、 Smoothing 、 Dynamic programming 、 Stereo cameras 、 MMX 、 Computational complexity theory 、 Computer science 、 Application software 、 Computation
摘要: A Dynamic Programming (DP) based algorithm that achieves real-time, high quality stereo-matching is presented. special coarse to fine approach as well the MMX assembler implementation mainly contribute reached computation speed. novel vertical smoothing by re-integrating paths inside DP matching allows avoidance of noisy horizontal strokes, so achieved. The current capable running at about 30 FPS on an 2.2GHz PC, which sufficient for utilization in real-time applications.