Toward improved ranking metrics
- 1 October 2000
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 22 (10) , 1132-1143
- https://doi.org/10.1109/34.879793
Abstract
In many computer vision algorithms, a metric or similarity measure is used to determine the distance between two features. The Euclidean or SSD (sum of the squared differences) metric is prevalent and justified from a maximum likelihood perspective when the additive noise distribution is Gaussian. Based on real noise distributions measured from international test sets, we have found that the Gaussian noise distribution assumption is often invalid. This implies that other metrics, which have distributions closer to the real noise distribution, should be used. In this paper, we consider three different applications: content-based retrieval in image databases, stereo matching, and motion tracking. In each of them, we experiment with different modeling functions for the noise distribution and compute the accuracy of the methods using the corresponding distance measures. In our experiments, we compared the SSD metric, the SAD (sum of the absolute differences) metric, the Cauchy metric, and the Kullback relative information. For several algorithms from the research literature which used the SSD or SAD, we showed that greater accuracy could be obtained by using the Cauchy metric instead.Keywords
This publication has 24 references indexed in Scilit:
- CANDID: comparison algorithm for navigating digital image databasesPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2002
- Using surface model to correct and fit disparity data in stereo visionPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2002
- Efficient stereo with multiple windowingPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2002
- Quality measures for interactive image retrieval with a performance evaluation of two 3×3 texel-based methodsPublished by Springer Nature ,1997
- Efficiency issues related to probability density function comparisonPublished by SPIE-Intl Soc Optical Eng ,1996
- Efficient content-based image retrieval in digital picture collections using projections: (near)-copy locationPublished by Institute of Electrical and Electronics Engineers (IEEE) ,1996
- Efficient color histogram indexing for quadratic form distance functionsPublished by Institute of Electrical and Electronics Engineers (IEEE) ,1995
- Query by image example: the comparison algorithm for navigating digital image databases (CANDID) approachPublished by SPIE-Intl Soc Optical Eng ,1995
- Computational Experiments with a Feature Based Stereo AlgorithmPublished by Institute of Electrical and Electronics Engineers (IEEE) ,1985
- Robust StatisticsPublished by Wiley ,1981