Multi-View Euclidean Reconstruction
摘要
Given the pixel coordinates of the projections of 3D points in multiple images, we can find the 3D points and the camera matrices that satisfy the projection equations. However, without further constraints, the reconstruction will be defined up to an arbitrary projectivity. If the intrinsic parameters are known, we can achieve a Euclidean reconstruction that differs from the true one by a similitude. This is the most relevant case in practice, and we will focus on it in this chapter.