A Triangulation Method in 3D Reconstruction from Image Sequences

There are several approaches invented for solving the triangulation problem in computer vision. Most of them are not projective-invariant, which is very important in 3D reconstruction. Therefore, we propose the method in this paper to solve the problem. Instead of solving high order equation, we introduce second-order Lagrange multiplier. We get the result by iterative, and therefore the result will not be trapped in some local minima. We show extensive results on synthetic data to validate our algorithm.