This motion detection model consists of a two-dimensional grid of elementary motion detection circuits, each of which obeys simple processing rules yet together are surprisingly effective at detecting and quantifying motion even in very complex scenes.


bird1movie                       bird1motion                   bird13d_2

1.                                                        2.                                                                            3.


  1. The original motion input.
  2. The motion detected.  Intensity represents strength of the output and color represents the direction.
  3. A three-dimensional representation of the strength of the output (direction is not indicated).


The mathematical underpinnings of the model are described in Zanker (1996).  We have implemented Zanker’s model using MATLAB 7.0.  The elementary motion detection units of the model are correlation detectors (often referred to as “Reichardt detectors”).  They function by comparing the spatial distribution in light intensity of scenes displaced in time.  Correlated shifts in time and space are interpreted as motion.

This first example shows how the landing and subsequent hopping of a bird against a simple, stationary background stimulates the detector.