Abstract
To compensate for the inherent limitations of least square based surface fitting (regression) methods, in recent years robust statistical techniques have been developed, the results of which remain trustworthy even if the contaminating noise has a non-Gaussian distribution and/or part of the data is missing (for reviews see the books of Huber, 1981; Rousseeuw and Leroy, 1987)
© 1989 Optical Society of America
PDF ArticleMore Like This
C. Henry Chu and Edward J. Delp
ThA1 Signal Recovery and Synthesis (SRS) 1989
Abdallah K. Cherri and Mohammad A. Karim
MH3 Optical Computing (IP) 1989
Charles S. Stein, Andrew B. Watson, and Lewis E. Hitchner
ThA7 Applied Vision (AV) 1989