很详细的单应矩阵分解R、t过程
附赠自动驾驶学习资料和量产经验：链接
已有多种方法将单应矩阵H分解为R、t，在《Deeper understanding of the homography decomposition for vision-based control》一文中介绍了三种方法：
O. Faugeras and F. Lustman, “Motion and structure from motion in a piecewise planar environment”, 1988.
Z. Zhang and A.R. Hanson, “3D Reconstruction based on homography mapping”, 1996.
E. Malis and M. Vargas, “Deeper understanding of the homography decomposition for vision-based control”, 2007.
以下仅介绍Faugeras SVD-based decomposition
