Performance Assessment by Resampling:
Rigid Motion Estimators
Bogdan Matei(1) Peter Meer(1) David Tyler(2)
(1)Department of Electrical and Computer Engineering
(2)Department of Statistics
Rutgers University, Piscataway, NJ 08855, USA
Quantitative assessment of performance in image understanding tasks
with real
data is difficult since the data is complex and the different
computational
modules most often interact. Employing modern statistical techniques
we have
developed a set of numerical tools which provide rigorous performance
measures
derived solely from the given input. Covariance matrices and
confidence
intervals are computed for the estimated parameters and individually
for the
corrected data points. As an example, the proposed methodology is
applied to
compare rigid motion estimators.
Appeared in
Empirical Evaluation Techniques in Computer Vision ,
K.W. Bowyer, P.J. Phillips (Eds.), IEEE CS Press, Los Alamitos, CA,
1998, 72--95.
Return to Research: Bootstrap as a tool for computer vision
