Computational and Optical Light Field Processing

在过去的几年里，出现了大量新的相机和显示技术，这些技术将从根本上改变我们捕捉物理世界和与之互动的方式。在多伦多大学，我们建议建立一个实验设施，使人们能够探索计算机视觉和摄影应用中迄今尚未探索的大型计算相机空间。这些技术包括(1)以高精度和分辨率以视频速率捕捉人、物体和环境的几何形状的新一代“深度相机”；(2)能够成像不同物体材料和外观的3D扫描仪-从漫反射的、有光泽的、到透明的；(3)同时捕捉一堆高分辨率、不同焦点的照片的“焦叠式相机”；以及(4)使用这种成像系统进行物体识别、跟踪和基于内容的图像检索的计算机视觉算法。

The last few years have ushered a wealth of new camera and display technologies that are poised to fundamentally change the ways we capture and interact with the physical world. At the University of Toronto, we propose to build an experimental facility that enables exploration of a large, hitherto-unexplored space of computational cameras for computer vision and photographic applications. These include (1) next-generation "depth cameras" for capturing the geometry of people, objects and environments at video rates, with high accuracy and resolution; (2) 3D scanners capable of imaging diverse object materials and appearances---from diffuse, to shiny, to transparent; (3) "focal-stack cameras" that capture a stack of high-resolution, differently-focused photos simultaneously; and (4) computer vision algorithms that use such imaging systems for object recognition, tracking, and content-based image retrieval.