Presidential Young Investigator Award (Computer and Information Science)

该项目专注于通过环境导航和操纵对象等任务的视觉运动分析。这项工作将基于过去五年的研究，解决运动测量和从相对运动中恢复三维(3-D)结构的问题。未来的工作将首先研究三个基本问题：第一，提取3D结构和运动的定性属性，以促进在环境中的快速运动；第二，扩展现有的从运动到结构的方法，以允许解释更大范围的非刚性运动；第三，设计并行网络来恢复三维结构，这将有助于使用自然图像运动序列来测试算法。这项研究的主旨将继续结合对这些问题的理论分析，算法的设计和测试，以及关于人类视觉系统运动分析的观察。

This project focusses on the analysis of visual motion for tasks such as navigation through the environment and manipulation of objects. The work will be based on previous research over the past five years, addressing the problems of motion measurement and recovery of three-dimensional (3-D) structure from relative motion. Future efforts will first examine three fundamental problems: First, the extraction of qualitative properties of 3-D structure and motion to facilitate fast locomotion through the environment; second, the extension of current structure-from- motion methods to allow the interpretation of a greater range of nonrigid motions, and third, the design of parallel networks to recover 3-D structure, which will facilitate the testing of algorithms with natural image motion sequences. The thrust of this research will continue to combine theoretical analyses of these problems, the design and testing of algorithms, and observations regarding the analysis of motion by the human visual system.