Please choose your delivery country and your customer group
This work of this proposal was concerned with the multidisciplinary field controlled active vision which involves the synthesis of techniques from control and computer vision to treat a number of fundamental problems including visual tracking. A key theme of our research was the development of techniques for using visual information in feedback control systems. Controlled active vision is leading to enhanced man-machine interfaces for interactions with computers and more complicated systems such as remote controlled weapons and vehicles. Our work has drawn upon our extensive experience in robust control, and the methods we have been developing for various problems in image processing and computer vision utilizing the theory of geometric variational evolution equations. These techniques have already been applied to visual tracking, automatic target recognition, and problems in biomedical engineering including image-guided surgery. It is important to note that many of these methods were derived from ideas in optimal control. In particular, the geometric variational techniques have been very influenced by concepts from optimal control, and the resulting concept of 'viscosity solution' is a direct consequence of Hamilton- Jacobi theory. 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 biological and machine vision. These problems all become manifest when one attempts to use a visual sensor in an uncertain environment, and to feed back in some manner the information.