Highly Efficient Surface Normal Integration

Abstract

The integration of surface normals for the computation of a surface in 3D space is a classic problem in computer vision. However, even nowadays it is still a challenging task to device a method that combines the flexibility to deal with non-trivial computational domains with high accuracy, robustness and computational efficiency. In this paper we propose to use for the first time in the literature Krylov subspace solvers as a main step in tackling the task. While these methods can be very efficient, they may only show their full potential when combined with a numerical preconditioning and even more importantly, a suitable initialization. To address the latter issue we propose to compute this initial state via a recently developed fast marching integrator. Numerical experiments prove the benefits of this novel combination.