Numerical Analysis of Nonlinear Equations in Computer Vision and Robotics
The need to apply sophisticated numerical analysis algorithms to computer vision and robotics problems is urgent, and these fields provide unique challenges different from the engineering problems normally encountered by numerical analysts. The opportunities for crossfertilization are vast; within computer vision, facet modelling, surface approximation, three-dimensional object recognition, and range data analysis require splines, generalized polynomials, classical approximation theory, approximation in various norms, robust statistics, and fixed point theory. Fundamental vision problems such as shape from shading, structure from motion, consistent labelling, and surface segmentation involve nonlinear equations, nonlinear optimization, quasi-Newton and homotopy algorithms Robot control, kinematics, and planning problems involve. modern differential and algebraic geometry, modern control theory, computational geometry, and homotopy theory.
Unable to display preview. Download preview PDF.
- A. Mitiche, S. Seida, and J. K. Aggarwal, Line-based computation of structure and motion using angular invariance, in Proc. Workshop on Motion: Representation and Analysis, Kiawah Island, SC, May, 1986, IEEE Computer Society Press.Google Scholar
- A. P. Morgan, A. Sommese, and L. T. Watson, Finding all solutions to polynomial systems using HOMPACK, General Motors Technical Report GMR-6109, GM Research Labs, Warren, MI 48090, 1988.Google Scholar
- R. M. Haralick, From facets to regions, manuscript, private communication, 1985.Google Scholar