STTR Phase I: Eye-Gaze Correction for more Effective Telepresence

这个小型企业技术转移研究（STTR）项目将开发一种新型的实时深度传感器，该传感器利用平方反比定律，例如，光辐射随距离的变化而下降。与现有的商用全画幅传感器相比，它有望提供高分辨率（例如640x480）深度图，同时将成本降低一到两个数量级。其次，它采用混合方法进行场景深度估计，在基于优化的概率框架中结合主动传感器和被动立体视觉。预计项目的结果将显著提高合成视图的质量和健壮性。网真市场正以100%的年增长率增长，预计2011年市场规模将达到10亿美元。预计所提出的技术可以为远程呈现系统提供巨大的通信改进。随着摄像头、大型显示器和宽带变得越来越普遍，远程呈现可以在每一台台式机和笔记本电脑上实现。有了正确的眼睛凝视、真人大小的图像和流畅的运动，引人注目的远程呈现体验有一天会大大减少对旅行的需求。此外，提出的主动距离传感器可以提供具有深度信息的视频流，可用于许多市场，如沉浸式游戏，智能汽车，安全，远程教育和工厂自动化。

This Small Business Technology Transfer Research (STTR) project will develop a novel real-time depth sensor that exploits the inverse square law, e.g., the light radiance falls off as a function of distance. Compared to existing commercial full-frame sensors, it is expected to provide high resolution (e.g. 640x480) depth maps while reducing the cost by one or two orders of magnitude. Secondly it utilizes a hybrid approach for scene depth estimation, combing both active sensors and passive stereovision in an optimization-based probabilistic framework. The outcome from the project is expected to significantly improve the quality and robustness of the synthesized view.The telepresence market is enjoying 100% annualized growth with a projected market size of $1B in year 2011. It is anticipated that the proposed technology could offer tremendous communication improvements for telepresence systems. As cameras, large displays, and broadband become more and more ubiquitous, telepresence can be realized on every desktop and laptop. With correct eye gaze, life size imagery, and fluid motion, compelling telepresence experiences could someday significantly reduce the need for travel. Additionally, the proposed active range sensor can deliver video streams with depth information, which can be used in many markets such as immersive gaming, intelligent automotive, security, distance education, and factory automation.