1st International ICST Workshop on Computing and Communications from Biological Systems: Theory and Applications

Research Article

Retina-Inspired Visual Processing

Download477 downloads
  • @INPROCEEDINGS{10.4108/ICST.BIONETICS2007.2418,
        author={Tyler W. Garaas and Marc Pomplun},
        title={Retina-Inspired Visual Processing},
        proceedings={1st International ICST Workshop on Computing and Communications from Biological Systems: Theory and Applications},
        proceedings_a={CCBS},
        year={2008},
        month={8},
        keywords={Biological Vision   Computer Vision   Motion Analysis   Retina Simulation},
        doi={10.4108/ICST.BIONETICS2007.2418}
    }
    
  • Tyler W. Garaas
    Marc Pomplun
    Year: 2008
    Retina-Inspired Visual Processing
    CCBS
    IEEE
    DOI: 10.4108/ICST.BIONETICS2007.2418
Tyler W. Garaas1,*, Marc Pomplun1,*
  • 1: Visual Attention Laboratory Computer Science Department University of Massachusetts Boston 100 Morrissey Boulevard Boston, MA 02114, USA
*Contact email: tgaraas@cs.umb.edu, marc@cs.umb.edu

Abstract

The processing that occurs in an animal’s retina is much more complex than once believed. In addition to highlighting high-contrast areas, the retina also performs a simple motion analysis of the visual field. Moreover, a full understanding of the neural functioning that occurs within the retina is not likely to take place anytime soon. Nevertheless, our present knowledge of the organization of retinal neurons and their responses to stimulation is enough to model meaningful and effective visual processing systems after. In this paper, following a high-level overview of retinal organization and functioning, we present a neural system that mimics the simple motion analysis that occurs within the retina to demonstrate the potential of retinal modeling. The neural system described is modeled after a subset of the connections that are present in most vertebrate retinas. Retina-inspired models such as this one could perform the initial processing step in a much more encompassing bio-inspired computer vision system.