Multi-modal primitives as functional models of hyper-columns and their use for contextual integration

Abstract

Computational Neuroscience, University of Stirling. In this paper, we describe a biological motivated image representalion in terms of local multi-modal primitives. These primitives are functional abstractions of hypercolumns in V1 [13], The efficient and generic coding of visual information in terms of local symbolic descriptiones allows for a wide range of applications. For example, they have been used to investigate the multi-modal character of Gestalt laws in natural scenes [14], to code a multi-modal stereo matching and to investigate the role of different visual modalities for stereo [11], and to use a combination of stereo and grouping as well as Rigid Body Motion to acquire reliable 3D information as demonstrated in this publication. © Springer-Verlag Berlin Heidelberg 2005.