Quantitative assessment of performance in image understanding tasks with real data is di cult since the data is complex and the di erent 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 con dence 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. 1: Performance assessment in image understanding The lack of universally accepted, rigorous performance assessment methodology is considered by many as one of the major bottlenecks of progress in image understanding. In a recent paper Christensen and Forstner [7] discuss several objections against the widespread use of evaluation techniques. Most of these objections are well justi ed. Performance for real data often cannot be reliably predicted from the controlled experiments with simple, synthetic inputs.
|
1034
|
An Introduction to the Bootstrap
– Efron, Tibshirani
- 1993
|
|
684
|
Robust regression and outlier detection
– ROUSSEEUW, LEROY
- 1987
|
|
373
|
Closed-form solution of absolute orientation using unit quaternions
– Horn
- 1987
|
|
236
|
LeastSquares Fitting of Two 3-D Point Sets
– Arun, Huang, et al.
- 1987
|
|
181
|
Statistical Optimization for Geometric Computation: Theory and Practice
– Kanatani
- 1996
|
|
135
|
Geometric Computation for Machine Vision
– Kanatani
- 1995
|
|
112
|
Subspace methods for recovering rigid motion i: algorithm and implementation
– Jepson, Heeger
- 1992
|
|
105
|
Least-squares estimation of transformation parameters between two point patterns
– Umeyama
- 1991
|
|
71
|
3D Dynamic Scene Analysis
– Zhang, Faugeras
- 1992
|
|
62
|
A leisurely look at the bootstrap, the jackknife and cross validation
– Efron, Gong
- 1983
|
|
61
|
Bootstrap methods for standard errors, confidence intervals, and other measures of statistical accuracy, Statistical Science l
– Efron, Tibshirani
- 1986
|
|
58
|
Measurement Error in Nonlinear Models
– Carroll, Ruppert, et al.
- 1995
|
|
58
|
Estimating 3-d rigid body transformations: a comparison of four major algorithms
– EGGERT, LORUSSO, et al.
- 1997
|
|
53
|
Motion and Structure from Image Sequences
– Weng, Huang, et al.
- 1993
|
|
39
|
Analysis of 3-D rotation fitting
– Kanatani
- 1994
|
|
36
|
Unmasking Multivariate Outliers and Leverage Points (with discussion
– Rousseeuw, Zomeren
- 1990
|
|
33
|
Relative orientation revisited
– Horn
- 1991
|
|
32
|
Jackknife, bootstrap and other resampling methods in regression analysis
– Wu
- 1986
|
|
31
|
Performance assessment through bootstrap
– Cho, Meer, et al.
- 1997
|
|
17
|
Least-squares tting of two 3-D point sets
– Arun, Huang, et al.
- 1987
|
|
15
|
Estimation with bilinear constraints in computer vision
– Leedan, Meer
- 1998
|
|
14
|
On the estimation of rigid body rotation from noisy data
– Goryn, Hein
- 1995
|
|
14
|
A new closed-form solution for absolute orientation
– Wang, Jepson
- 1994
|
|
11
|
Performance assessment of near-perfect machines
– Haralick
- 1989
|
|
9
|
Robust Statistics: The Approach Based on
– Hampel, Ronchetti, et al.
- 1986
|
|
7
|
Statistics: With a View Toward Applications
– Breiman
- 1973
|
|
6
|
Analysis of 3-D rotation tting
– Kanatani
- 1994
|
|
5
|
Performance characteristics of vision algorithms
– Christensen, Forstner
- 1997
|
|
5
|
Bootstrap: More than a stab in the dark
– Young
- 1994
|
|
3
|
Bootstrap methodology
– Babu, Rao
- 1993
|
|
3
|
Some asymptotic theory for the bootstrap,"Annals
– Bickel, Freedman
- 1981
|
|
2
|
The bootstrap - A review
– Manteiga, Sanchez, et al.
- 1994
|
|
1
|
Statistics: With a View Toward
– Breiman
- 1973
|
|
1
|
open c n:P | n ‘a
– unknown authors
- 1997
|
|
1
|
Jackknife variance estimation and bias reduction
– Peddada
- 1993
|
|
1
|
Zomeren, "Unmasking multivariate outliers and leverage Points
– Rousseeuw, van
- 1990
|
|
1
|
Robust covariance estimates based on resampling
– Stromberg
- 1997
|
|
1
|
The standardized in uence matrix and its applications
– Lu, Ko, et al.
- 1997
|
