Geometric Variational Methods for Controlled Active Vision

摘要

The key objective of this project is the use of visual information in a feedback loop, the underlying problem of controlled active vision. The problem of controlled active vision, and in particular visual tracking requires the integration of techniques from control theory, signal processing, and computer vision. For some time now the role of control theory in vision has been recognized. In particular, the branches of control that deal with system uncertainty, namely adaptive and robust, have been proposed as essential tools in coming to grips with the problems of both machine and biological vision. Visual tracking provides a fundamental example of the need for controlled active vision. While tracking in the presence of a disturbance is a classical control problem, visual tracking raises new issues. First since cameras are part of the system, one must consider the nature of the disturbance from imaging sensors. The feedback signal may require some interpretation of the image, for example segmentation of a target from its background, or an inference about an occluder. In the project, we expressly emphasize active vision, because the result may be viewpoint dependent. In particular, calibration may influence the control law. And finally, as visual processing becomes more complex, the issue of processing time arises. Each of these problems must be answered before target detection, and visually-mediated control can be provided for advanced weapon systems.