View-Consistent 4D Light Field Superpixel Segmentation

Many 4D light field processing methods and applications rely on superpixel segmentation, for which occlusion-aware view consistency is important. Yet, existing methods often enforce consistency by propagating clusters from a central view only, which can lead to inconsistent superpixels for non-central views. Our proposed approach combines an occlusion-aware angular segmentation in horizontal and vertical epipolar plane image (EPI) spaces with a clustering and propagation step across all views. Qualitative video demonstrations show that this helps to remove flickering and inconsistent boundary shapes versus the state-of-the-art light field superpixel approach (LFSP [25]), and quantitative metrics reflect these findings with greater self similarity and fewer numbers of labels per view-dependent pixel.