This paper empirically compares nine families of image dissimilarity measures that are based on distributions of color and texture features summarizing over 1000 CPU hours of computational experiments. Ground truth is collected via a novel random sampling scheme for color, and by an image partitioni
Characterization of empirical discrepancy evaluation measures
✍ Scribed by N.L Fernández-Garcı́a; R Medina-Carnicer; A Carmona-Poyato; F.J Madrid-Cuevas; M Prieto-Villegas
- Publisher
- Elsevier Science
- Year
- 2004
- Tongue
- English
- Weight
- 878 KB
- Volume
- 25
- Category
- Article
- ISSN
- 0167-8655
No coin nor oath required. For personal study only.
📜 SIMILAR VOLUMES
## Abstract For Abstract see ChemInform Abstract in Full Text.
Two possible measures of hardness are proposed. An average chemical hardness is calculated separately for acidic and basic reactions of atoms. Differential hardnesses are derived from atomic radii.
## a b s t r a c t In this work, we establish a moderate deviation principle for the L 1 -error of empirical measures on partitions. As an application, the moderate deviation principle for the L 1 -error of the histogram density estimator is also obtained.