Abstract

Letter to the editor Dear Editor, In a recent paper Balboa, Tyler, and Grzywacz (2001) report findingswhich contradict elementsof my work on the originsof scaling in natural images(Ruderman, 1997). Here I refute their claim and in addition show that the theoretical approximationspresented in their Appendix are not appropriate for ensembles of natural images. In a self-similar image ensemble any ensemble averaged statistic has a spatial dependence which is pure power-law. A change in scale alters the coefficient in front of thispower-law term but doesnot change its functional form. One such statistic is the ensemble power spectrum which has been shown to obey a powerlaw for over 2.5 decadesin spatial frequency (Ruderman