Variational Depth from Focus Reconstruction

TL;DR

This paper introduces a variational approach to depth from focus reconstruction, combining a nonconvex data fidelity term with convex regularization, solved efficiently via a linearized ADMM method, and validated on simulated and real data.

Contribution

It formulates depth from focus as a variational problem with a novel nonconvex data term and convex regularization, solved efficiently with a linearized ADMM approach.

Findings

The method produces more realistic depth maps.

It demonstrates robustness to noise.

It outperforms classical methods on tested datasets.

Abstract

This paper deals with the problem of reconstructing a depth map from a sequence of differently focused images, also known as depth from focus or shape from focus. We propose to state the depth from focus problem as a variational problem including a smooth but nonconvex data fidelity term, and a convex nonsmooth regularization, which makes the method robust to noise and leads to more realistic depth maps. Additionally, we propose to solve the nonconvex minimization problem with a linearized alternating directions method of multipliers (ADMM), allowing to minimize the energy very efficiently. A numerical comparison to classical methods on simulated as well as on real data is presented.