Looking Beyond the Camera Aperture: Learning from the Field of Light

Abstract

Recent imaging technologies have been rapidly evolving for sampling richer and more immersive representations of the 3D world. Compared with conventional cameras which only capture 2D projections of the world, the information embedded within the field of light rays opens extraordinary possibilities, which provides unique advantages in applications such as 3D Modelling, VR, and Immersive tele-presence.

In this seminar, I will introduce our work on applying data-driven, learning-based models for compressed sensing and smart interpretation of the high dimensional visual data. In particular, I will explain our ideas in statistical reasoning of the scene geometry, and how deep learning models can be designed to efficiently extract essential clues to facilitate operations such as 4D denoising, and spatial-angular super-resolution, which enable us to look freely from within, and even beyond the limit of camera base-lines.

Biography

Dr. CHEN Jie is currently a senior algorithm engineer at OmniVision Technologies, Inc. He was a post-doctoral research fellow at ST Engineering-NTU Corporate Laboratory, Singapore between 2016~2019. He received the B.S. and the M.Eng degree both from School of Optical and Electronic Information, Huazhong University of Science and Technology, China, and the Ph.D. degree in the School of Electrical & Electronic Engineering, Nanyang Technological University, Singapore, in 2016. His research interests are in Computational photography (systems and algorithms for light field/hyperspectral signal acquisition, compressed compressing and compression); Computational and machine learning methods for image/video restoration (denoising, vision enhancement, super resolution, motion analysis), 3D content modeling (stereo, view synthesis/rendering, baseline magnification). He is the author/co-author of more than 30 peer-reviewed journal and conference proceedings.