A multi-scale anti-aliasing rendering algorithm for the light field imaging

As aliasing artifacts can significantly reduce the quality of light field imaging, we propose a multi-scale anti-aliasing rendering algorithm to alleviate these artifacts. First, we explain that the angular under-sampling is an essential cause of aliasing artifacts. Then, we model the relationship between the aliasing and angular under-sampling. Distinct from existing methods, we address angular aliasing in the light field rendering stage, and the proposed algorithm does not need scene depth as the rendering prior. To avoid boundary problem, we introduce a multi-scale gradient field fusion algorithm, which can seam different level of non-aliased image parts together to reduce the light field aliasing. We test the proposed algorithm on both synthetic data and real scene data sets. In the experiments, the proposed rendering approach can significantly outperform the traditional light field rendering and can obtain similar results as the depth prior based rendering method.