Movement Detectors of the Correlation Type Provide Sufficient Information for Local Computation of the 2-D Velocity Field

Abstract

The projection of the velocity vectors of objects moving in three-dimensional space on the image plane of an eye or a camera can be described in terms of a vector field. This so-called 2-D velocity field is time-dependent and assigns the direction and magnitude of a velocity vector to each point in the image plane. The 2-D velocity field, however, is a purely geometrical concept and does not directly represent the input site of a visual information processing system. The only information available to a visual system is given by the time-dependent brightness values as sensed in the image plane by photoreceptors or their technical equivalents. From spatio-temporal coherences in these changing brightness patterns motion information is computed. This poses the question about whether the spatio-temporal brightness distributions contain sufficient information to calculate the correct 2-D velocity field. Here we show that the 2-D velocity field generated by motion parallel to the image plane can be computed by purely local mechanisms.