Iterative reconstruction of scene depth with fidelity based on light field data.

摘要: Depth reconstruction from the light field, as a depth extracting approach, is vibrant research field in computational imaging. In this paper, was approached an optimization problem by analyzing forward and inverse models. The models present connection between of scene. We proposed iterative method for scene with fidelity 4D data. objective function entails three terms, which matching term used term, while gradient classification constitute penalty terms. minimized correcting mismatching pixels iteratively. way, accuracy significantly improved high robustness, especially weak texture occlusion regions. experimental results show that map can be reconstructed method, alignment resolution An accurate 3D surface view achieved using information field.