A neural network for the detection of rotational motion
Abstract
Describes a biologically plausible neural network which detects rotational motion in the frontal-parallel plane (revolving around the Z-axis) produced by a multiple-dots stimulus against a stationary background. The network consists of multilayered velocity-sensitive sensory cells organized in a locally connected fashion. The network displays the ability to segment multiple rotational stimuli in time sequences as well as an object with both rotational and translational motion by extracting the center and the perimeter trace of the rotational stimulus. Different simulations have been performed, and the results are discussed.>
Authors: Irwin King, Jim‐Shih Liaw, Michael A. Arbib
Published in: International Joint Conference on Neural Networks (IJCNN) (2002)