Computational and Optical Processing Architectures for next-Generation Mobile Cameras

点击翻译按钮获取中文摘要

Mobile devices are fuelling a major shift in how we interact with each other and with the world around us. They are poised to become an extension of ourselves, with drastically-increased capabilities for sensing the physical world. Cameras are the "eyes" of mobile devices, and continuous improvements in their optical quality, speed and wireless networking capabilities have spawned new markets and new uses (e.g., image-based search, and assistive technologies for the visually impaired). While past work in mobile cameras focused primarily on making them "see better" (finer detail, more images per second, requiring less light), it is now time to ask "what else can mobile cameras see". The proposed work will develop and practically demonstrate the applications possible with a radically-different camera that has never-seen-before abilities. Not only will our camera technology image the world in three dimensions (i.e., the distance and 3D shape of objects it "sees") at a detail that far surpasses existing solutions, but it can take into account the 3D path that light followed to reach it. Our camera technology is based on an novel imaging technique called Primal-Dual Coding pioneered by the PIs, and enables numerous applications in 3D scanning, visual recognition, biometrics, inspection and scientific imaging. We have already demonstrated the potential of the underlying principle of operation using off-the-shelf components. With this project we seek to unlock our camera technology's the true potential by (1) fabricating custom CMOS imagers, (2) designing and building new mobile camera platforms, and (3) developing the associated hardware and software architectures to support them.