The usual algorithms in Computer Vision Structure and Motion are large non convex problems, often involving large numbers of variables (in excess of a million). Through a variety of simple techniques, it is possible to find initial solutions that will serve for initialization of iterative algorithms with good results. Nevertheless, it is interesting to find algorithms that are provably optimum, finding a guaranteed global minimum. This talk gives a summary of work in this area, involving a variety of techniques, including L-infinity optimization, branch and bound and convex verification techniques.
Richard Hartley received the BSc degree from the Australian National University (ANU) in 1971, the MSc degree in computer science from Stanford University in 1972, and the PhD degree in mathematics from the University of Toronto, Canada, in 1976. He is currently a professor and member of the computer vision group in the Department of Information Engineering at ANU. He also belongs to the Vision Science Technology and Applications Program in National ICT Australia, a government-funded research institute. He did his PhD thesis in knot theory and worked in this area for several years before joining the General Electric (GE) Research and Development Center, where he worked from 1985 to 2001. During the period 1985-1988, he was involved in the design and implementation of computer-aided design tools for electronic design and created a very successful design system called the Parsifal Silicon Compiler, described in his book Digit Serial Computation. In 1991, he was awarded GE’s Dushman Award for this work. Around 1990, he developed an interest in computer vision, and in 2000, he coauthored (with Andrew Zisserman) a book on multiple-view geometry. He has authored more than 100 papers in knot theory, geometric voting theory, computational geometry, computer-aided design, and computer vision and holds 32 US patents. He is a member of the IEEE.