Traditional cameras transform 3D scenes into 2D images so computer vision has become an ill-posed problem. So it is better to provide more visual information to support advanced computer vision. This project proposes to develop a novel camera to capture 4D light field information of visual scenes. Microlens array is used to record the direction information of light fields so that it is possible to reconstruct depth information of visual objects. In addition, image refocusing algorithms are developed to extend the depth of field of cameras. The innovations of light field cameras will significantly facilitate the research of computer vision. The main applications of light field cameras include iris and face recognition at a distance, visual surveillance, etc.
(Sources of Funding: Chinese Academy of Sciences)