U.S.-France Cooperative Research (INRIA): Detection and Characterization of Temporal Changes in Satellite Image Sequences

这项为期三年的美国-法国计算机视觉合作研究的奖项包括南加州大学机器人和智能系统研究所的Isaac Cohen，南加州大学和喷气推进实验室的Roberto Manduchi，以及法国rocqucourt信息学和应用数学研究所（INRIA）的空中项目主任Isabelle Herlin。美国和法国的研究小组将解决通过卫星遥感获得的环境图像序列的时间变化的检测和表征问题。所讨论的环境问题涉及植被和林业监测以及海洋/大气科学。计算机视觉技术被广泛用于检测和表征图像的时间变化。该提案将这项工作扩展到植被和森林图像中纹理变化检测的图像特性表征，以及云和海洋-大气交换等非刚性运动现象的物理特性确定。美国研究人员为这次合作带来了光流、变分方法和纹理建模方面的专业知识。这是法国团队在非刚性运动方面的专业知识的补充，特别是云的运动。该项目还利用了法国小组提供的陆地卫星遥感图像和地面真相测量。该项目将促进对卫星图像自动分析方法和技术的理解，并导致用于地球系统科学研究的新型计算机视觉工具的开发。该奖项代表了美国方面向美国国家科学基金会和法国国家信息学和应用数学研究所（INRIA）联合提出的一项建议。美国国家科学基金会将支付美国研究人员和研究生的旅费和生活费。INRIA将支持法国研究人员访问美国。

0004494CohenThis three-year award for U.S.-France cooperative research in computer vision involves Isaac Cohen of the Institute for Robotics and Intelligent Systems at the University of Southern California (USC), Roberto Manduchi of USC and the Jet Propulsion Laboratory, and Isabelle Herlin, Director of the Projet Air at the Institute for Research in Informatics and Applied Mathematics (INRIA) at Rocquencourt, France. The US and French research teams will address the problem of detection and characterization of temporal change in environmental image sequences obtained through satellite remote sensing. The environmental problems in question involve vegetation and forestry monitoring, and ocean/atmospheric sciences.Computer vision techniques are widely used for detecting and characterizing temporal changes in images. This proposal extends this work to the characterization of image properties for texture change detection as found in vegetation and forest images, and to the determination of physical properties for non rigid motion phenomena such as clouds, and ocean-atmosphere exchanges. The US investigators bring to this collaboration expertise in optical flow, variational methods and texture modeling. This is complemented by the French team's expertise in non rigid motion in general and the motion of clouds in particular. The project also takes advantage of access to Landsat remote sensing images and ground truth measures provided by the French team. The project will advance understanding of methods and techniques for automatic analysis of satellite images and lead to development of new computer vision tools for use in earth systems science research This award represents the US side of a joint proposal to the NSF and the French National Institute for Research in Informatics and Applied Mathematics (INRIA). NSF will cover travel funds and living expenses for the US investigators and graduate student. INRIA will support the visits of French researchers to the United States.