Conventional stereoscopic displays suffer from vergence-accommodation conflict and cause visual fatigue. Integral-imaging-based displays resolve the problem by directly projecting the sub-aperture views of a light field into the eyes using a microlens array or a similar structure. However, such displays have an inherent trade-off between angular and spatial resolutions. In this paper, we propose a novel coded time-division multiplexing technique that projects encoded sub-aperture views to the eyes of a viewer with correct cues for vergence-accommodation reflex. Given sparse light field sub-aperture views, our pipeline can provide a perception of high-resolution refocused images with minimal aliasing by jointly optimizing the sub-aperture views for display and the coded aperture pattern. This is achieved via deep learning in an end-to-end fashion by simulating light transport and image formation with Fourier optics. To our knowledge, this work is among the first that optimize the light field display pipeline with deep learning. We verify our idea with objective image quality metrics (PSNR, SSIM, and LPIPS) and perform an extensive study on various customizable design variables in our display pipeline. Experimental results show that light fields displayed using the proposed technique indeed have higher quality than that of baseline display designs.

We propose a lightweight yet effective deep learning pipeline for light field synthesis from a single stereo image pair. Our pipeline consists of a convolutional network (CNN) that enforces a left-right consistency constraint on the light fields synthesized from left and right stereo views, a stage that merges light fields synthesized from left and right stereo views with a novel alpha blending technique, and a final refinement network using a unique 3D convolution operation. Our experiments quantitatively and qualitatively confirm the effectiveness and robustness of the proposed model, which performs favorably against state-of-the-art algorithms for light field synthesis from extremely sparse (only one, two, or four) views while using much fewer parameters.

To rectify fisheye distortion from a single image, we advance self-supervised learning strategies and propose a unique deep learning model of Fisheye GAN (FE-GAN). Our FE-GAN learns pixel-level distortion flow from sets of fisheye distorted images and distortion-free ones (but not requiring such correspondences), with unique cross-rotation and intra-warping consistency introduced. With such novel self-supervised learning techniques, our FEGAN is able to recover the distortion-free image directly from the single fisheye image input. Our experiments quantitatively and qualitative confirm the effectiveness and robustness of our proposed model, which performs favorably against recent GAN-based image translation models.