Binocular Disparity Energy after
Read, Parker and Cumming
This binocular disparity energy (BDE) model has linear spatial filters as in the
BDE_Gabor model, but with the modification that the
monocular outputs of each initial stage filter are rectified before being binocularly
combined as described by Read et al (2002). There are 4 variants based on those some of
thise shown in their Figure 8: tuned excitatory (TE),
tuned inhibitory (TI), near (NE), and far (FA) models.
- Read JCA, Parker AJ, Cumming BG (2002) A simple model accounts for the
response of disparity-tuned V1 neurons to anticorrelated images. Vis Neurosci 19:735-753.
- Ohzawa I, DeAngelis GA, Freeman RD (1990) Stereoscopic Depth Discrimination
in the Visual Cortex: Neurons Ideally Suited as Disparity Detectors. Science 239:1047-1051.
The visual stimulus is processed (convolved) by four linear Gabor
filters. The icons on the top row show x-t slices of the 3D filters, while the
bottom row shows x-y slices for each subunit. The signals from the left and right
(even filters) and fl2
filters) are also inverted, and these positive and negative filter outputs are
then half-wave rectified before being combined via addition or subtraction, and
half-squared. The four resulting signals are then added to generate
, the binocular disparity energy.
The raw signal (bde) 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.