At all Costs: A Comparison of Robust Cost Functions for Camera Correspondence Outliers
Authors:
- Kirk MacTavishKMKirk MacTavish
- Timothy D. BarfootTBTimothy D. Barfoot
- 1 June 2015
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 62-69
Abstract
Camera-based localization techniques must be robust to correspondence errors, i.e., when visual features (landmarks)are matched incorrectly. The two primary techniques to address this issue are RANSAC and robust M-estimation -- each more appropriate for different applications. This paper investigates the use of different robust cost functions for M-estimation to deal with correspondence outliers, and assesses their performance under varying degrees of data corruption. Experimental results show that using an aggressive red ascending cost function (e.g., Dynamic Covariance Scaling (DCS) or Geman-McClure (G-M)) best improves accuracy by excluding outliers almost entirely. Additionally, adjusting an error-scaling parameter for the robust cost function over the course of the optimization improves convergence with poor initial conditions.Keywords
This publication has 13 references indexed in Scilit:
- Towards a reliable SLAM back-end2013
- Robust map optimization using dynamic covariance scaling2013
- A novel mapping strategy based on neocortex model: Pre-liminary results by hierarchical temporal memory2012
- Switchable constraints for robust pose graph SLAM2012
- Evaluation of heterogeneous measurement outlier rejection schemes for robotic planetary surface mappingActa Astronautica, 2012
- A nonlinear filter for film restoration and other problems in image processingCVGIP: Graphical Models and Image Processing, 1992
- Random sample consensusCommunications of the ACM, 1981
- Robust regression using iteratively reweighted least-squaresCommunications in Statistics - Theory and Methods, 1977
- The Fitting of Power Series, Meaning Polynomials, Illustrated on Band-Spectroscopic DataTechnometrics, 1974
- Robust Estimation of a Location ParameterThe Annals of Mathematical Statistics, 1964