Motion Energy, Gabor Filter
This non-opponent motion energy (ME) model has linear filters that are
3-D Gabor functions (Gaussian times sinusoid). A 2-dimensional form
of this model was first described by Adelson and Bergen (1985; their
Fig 9). This type of ME unit has been used to model responses of
direction selective (DS) cortical neurons (Heeger, 1987; Grzywacz and
Yuille, 1990), although it fails to capture some basic properties of
DS neurons in macaque (Bair and Movshon, 2004).
- Adelson EH, Bergen JR (1985) Spatiotemporal energy models for the
perception of motion. J Opt Soc Am A 2:284-299.
- Bair W, Movshon JA (2004) Adaptive temporal integration of
motion in direction-selective cells in macaque visual cortex. J
- Grzywacz NM, Yuille AL (1990) A model for the estimate of
local image velocity by cells in the visual cortex. Proc Roy Soc Lond
- Heeger DJ (1987) Model for the extraction of image flow. J
Opt Soc Am A 4:1455-1471.
The visual stimulus is processed (convolved) by two linear Gabor
filters (icons show x-t slices of 3D filters) to yield the signals
(filter preferred even) and fpo
(filter preferred odd).
The filter outputs are squared, and the signals are added to form
, the motion energy in the preferred direction.
The raw signal (mep) is then offset, scaled and half-wave rectified
(although the signal is typically already non-zero unless the scaling
or offset has introduced negative values), and it is used to drive a
Poisson spiking mechanism. The spikes are time shifted to simulate a
neurobiological latency. See the model (.moo) files for the
parameters that govern these computations.