Scene dependence of the non-Gaussian scaling properties of natural images

Citation
A. Nevado et al., Scene dependence of the non-Gaussian scaling properties of natural images, NETWORK-COM, 11(2), 2000, pp. 131-152
Citations number
36
Categorie Soggetti
AI Robotics and Automatic Control
Journal title
NETWORK-COMPUTATION IN NEURAL SYSTEMS
ISSN journal
0954898X → ACNP
Volume
11
Issue
2
Year of publication
2000
Pages
131 - 152
Database
ISI
SICI code
0954-898X(200005)11:2<131:SDOTNS>2.0.ZU;2-J
Abstract
We report results on the scaling properties of changes in contrast of natur al images in different visual environments. This study confirms the existen ce, in a vast class of images, of a multiplicative process relating the var iations in contrast seen at two different scales, as was found in Turiel et al (Turiel A, Mate G, Parga N and Nadal J-P 1998 Self-Similarity Propertie s of Natural I,Images: Proc. NIPS '97(Cambridge, MA: MIT Press), Turiel A, Mate G, Parga N and Nadal J-P 1998 Phys. Rev. Lett. 80 1098-101). But it al so shows that the scaling exponents are not universal: even if most images follow the same type of statistics, they do it with different values of the distribution parameters. Motivated by these results, we also present the a nalysis of a generative model of images that reproduces those properties an d that has the correct power spectrum. Possible implications for visual pro cessing are also discussed.