psychopy_ext.models.HMAX.get_gaussians_matlab

HMAX.get_gaussians_matlab(filter_sizes_all, n_ori=4, sigDivisor=4.0)[source]

Generates 2D difference of Gaussians (DoG) filters. This is the original version of DoG filters used in HMAX. It was written in a very cumbersome way and thus I replaced it by the gaussian_filters function. If you want to produce identical numerical values of the filters, you should use this function. Otherwise, gaussian_filters does the job just as well, but much nicer.

Parameters
filter_sizes_all: list (depth 2))
A nested list (grouped by filter bands) of integer filter sizes
n_ori: int
A number of filter orientations (default: 4) Orientations are spaced by np.pi/n_ori
sigDivisor: float
A parameter to adjust DoG filter frequency (default: 4.)
Returns
gaussians: list (depth 2)
A nested list of filters of all orientations