Heteroscedastic Projection Based M-Estimators

Raghav Subbarao and Peter Meer

Department of Electrical and Computer Engineering
Rutgers University, Piscataway, NJ 08854, USA

Robust regression methods, such as RANSAC, suffer from a sensitivity to the scale parameter used for generating the inlier-outlier dichotomy. Projection based M-estimators (pbM) offer a solution to this by reframing the regression problem in a projection pursuit framework. In this paper we modify the pbM formulation to obtain an improved pbM algorithm. Furthermore, the modified algorithm is easily generalized to handle heteroscedastic data . The superior performance of heteroscedastic pbM, as compared to simple pbM, is experimentally verified.

Workshop on Empirical evaluation Methods in Computer Vision, San Diego, CA, June 2005 (in conjunction with CVPR'05 ).
Return to Research: Robust Analysis of Visual Data        Return to List of Publications
Download the paper