A buyer's guide to conic fitting
BMVC '95 Proceedings of the 6th British conference on Machine vision (Vol. 2)
Orthogonal least squares fitting by conic sections
Proceedings of the second international workshop on Recent advances in total least squares techniques and errors-in-variables modeling
On a Variational Formulation of the Generalized Singular Value Decomposition
SIAM Journal on Matrix Analysis and Applications
Direct Least Square Fitting of Ellipses
IEEE Transactions on Pattern Analysis and Machine Intelligence
On the Fitting of Surfaces to Data with Covariances
IEEE Transactions on Pattern Analysis and Machine Intelligence
FNS, CFNS and HEIV: A Unifying Approach
Journal of Mathematical Imaging and Vision
Functions of a Complex Variable: Theory and Technique (Classics in Applied Mathematics)
Functions of a Complex Variable: Theory and Technique (Classics in Applied Mathematics)
Performance evaluation of iterative geometric fitting algorithms
Computational Statistics & Data Analysis
Robust adaptive MMSE/DFE multiuser detection in multipath fading channel with impulse noise
IEEE Transactions on Signal Processing
| Hi-index | 0.01 |
This work explains the geometry of analytic theorems by Forsythe and Golub, Gander, and Moré to the effect that the constrained minimum (respectively, maximum) of a positive definite Hermitian form on a level set of any nonnecessarily definite Hermitian polynomial corresponds to the Lagrange multiplier with the smallest (respectively, largest) absolute value. Locally, the law of sines for the triangle with vertices at the center and at two stationary points reveals that the objective values are in the same order as the magnitudes of the Lagrange multipiers. Global geometry explains the same results for global minima and global maxima by showing that the constraining quadric surface and the Lagrange multiplier form a set of geodetic coordinates for the entire ambient space. Duality and the symmetry of the constraining quadratic hypersurface also explain why the difference between two stationary points with the same objective value is a generalized eigenvector.