Discontinuity-preserving Normal Integration with Auxiliary Edges
This work addresses a specific challenge in surface reconstruction for computer vision and graphics, offering an incremental improvement over existing discontinuity-preserving normal integration techniques.
The paper tackles the problem of reconstructing accurate depth maps from normal maps with discontinuities, such as those from self-occlusion, by introducing auxiliary edges to explicitly model hidden jumps, resulting in more accurate reconstruction of subtle discontinuities compared to previous methods.
Many surface reconstruction methods incorporate normal integration, which is a process to obtain a depth map from surface gradients. In this process, the input may represent a surface with discontinuities, e.g., due to self-occlusion. To reconstruct an accurate depth map from the input normal map, hidden surface gradients occurring from the jumps must be handled. To model these jumps correctly, we design a novel discretization scheme for the domain of normal integration. Our key idea is to introduce auxiliary edges, which bridge between piecewise-smooth patches in the domain so that the magnitude of hidden jumps can be explicitly expressed. Using the auxiliary edges, we design a novel algorithm to optimize the discontinuity and the depth map from the input normal map. Our method optimizes discontinuities by using a combination of iterative re-weighted least squares and iterative filtering of the jump magnitudes on auxiliary edges to provide strong sparsity regularization. Compared to previous discontinuity-preserving normal integration methods, which model the magnitudes of jumps only implicitly, our method reconstructs subtle discontinuities accurately thanks to our explicit representation of jumps allowing for strong sparsity regularization.