Title :
Algorithms for shape from shading and occluding boundaries
Author_Institution :
AT&T Bell Lab., Murray Hill, NJ, USA
Abstract :
The problem of shape from shading and occluding boundaries can be reduced to solving a system of nonlinear equations for a smoothing spline, using regularization. The author discusses some theoretical issues corrected with this problem. He studies the existence and uniqueness of the smoothing spline. For solving the system of equations, he presents an iterative algorithm and proves that it converges to the unique solution for arbitrary regions. He also discusses efficient implementations of the iterative method, and analyzes their computational complexity
Keywords :
computational complexity; computerised pattern recognition; splines (mathematics); computational complexity; computerised pattern recognition; existence; iterative algorithm; occluding boundaries; regularization; shading; shape; smoothing spline; uniqueness; Boundary conditions; Brightness; Computational complexity; Image converters; Iterative algorithms; Iterative methods; Linear systems; Nonlinear equations; Shape; Spline;
Conference_Titel :
Computer Vision and Pattern Recognition, 1988. Proceedings CVPR '88., Computer Society Conference on
Conference_Location :
Ann Arbor, MI
Print_ISBN :
0-8186-0862-5
DOI :
10.1109/CVPR.1988.196278