Author(s)

J. W. Zweck and L. R. Williams

Abstract

Euclidean invariant computation of stochastic completion fields using shiftable-twistable wavelets J. W. Zweck and L. R. Williams Department of Computer Science University of New Mexico, USA Abstract Computations in visual cortex have many features in common with compu- tations in the continuum, even though they are implemented in a discrete network. In particular they are Euclidean invariant: An arbitrary rotation and translation of the input produces an identical transformation of the output. We introduce the notion of a shiftable-twistable wavelet basis and show how it can be used to perform Euclidean invariant discrete computa- tions on the continuous space of positions and directions. The particular computation we consider is that of completing the boundaries of partially occluded objects. 1 Introduction Any computational model of human visual information processing must reconcile two apparently contradictory observations. First, com- putations in primary visual cortex

Keywords