Faculty Resources

A Two-Dimensional Visual Motion Detector Based on Biological Principles



This report describes a biologically-inspired computer model of visual motion detection.  The model is designed to use sequential digital frames (e.g. typically from digital video of natural scenes) and detect the local direction and relative strength of motion in any portion of the scene.  The aim of creating this model was to study the ways that changes in the properties of simple motion detection circuits alter the way in which natural moving stimuli are likely to be perceived.  The model is based on a 2-dimensional grid of “correlation-type” motion detector, frequently referred to as “Reichardt detectors.”