Robust Regression with Projection Based M-estimators

  • Authors:
  • Haifeng Chen;Peter Meer

  • Affiliations:
  • -;-

  • Venue:
  • ICCV '03 Proceedings of the Ninth IEEE International Conference on Computer Vision - Volume 2
  • Year:
  • 2003

Quantified Score

Hi-index 0.00

Visualization

Abstract

The robust regression techniques in the RANSAC family arepopular today in computer vision, but their performance dependson a user supplied threshold. We eliminate this draw-backof RANSAC by reformulating another robust method,the M-estimator, as a projection pursuit optimization problem.The projection based pbM-estimator automatically derivesthe threshold from univariate kernel density estimates.Nevertheless, the performance of the pbM-estimator equalsor exceeds that of RANSAC techniques tuned to the optimalthreshold, a value which is never available in practice.Experiments were performed both with synthetic and realdata in the affine motion and fundamental matrix estimationtasks.