This section has examined some speculative models advanced by Zorkoczy (1966), Reichardt (1964), and Fukushima (1969) that have attempted to describe certain aspects of feature extraction observed in vertebrate and insect visual systems. Zorkoczy's models are particularly interesting because he introduces the basis for spatiotemporal filters that respond to an object's shape as well as its velocity. The use of a delay between signals from adjacent receptors appears as a recurrent theme in many other visual receptor array models exhibiting DS. Indeed, Reichardt's basic neural correlator model (see Figure 7.1-8), used to describe insect optomotor responses not only uses delays, but also signal multiplication. (A review of DS models can be found in Kien, 1975.)
Fukushima's pattern recognition models work on static, nonmoving objects. They are based on the neural equivalents of Fourier optic image processors. Continuous equivalents of discrete weighting functions are convolved with signals in a sequence of planes to extract an object's features such as straight lines at some angle to a reference axis (such as vertical).
While these models may appear simple, they laid the groundwork for and were the basis of subsequent innovations in machine vision and visual signal processing by artificial neural networks (ANNs).
Was this article helpful?
This guide will help millions of people understand this condition so that they can take control of their lives and make informed decisions. The ebook covers information on a vast number of different types of neuropathy. In addition, it will be a useful resource for their families, caregivers, and health care providers.