Describes what the authors consider are fundamental building blocks for geometric computer vision or structure-from-motion: epipolar geometry, pose and motion estimation, 3D scene modeling, and bundle adjustment. The main goal is to highlight the core principles of these, which are independent of specific camera models.
