User-friendly constructor interface for the Conovler.
An array of images with which we convolve each input image. These images should *not* be pre-whitened.
Metadata of a typical image we will be convolving. All images must have the same size/shape.
Whitener to be applied to both the input images and the filters before convolving.
Should the patches be normalized before convolution?
Constant to be used in scaling.
Should the filters be flipped before convolution is applied (used for comparability to MATLAB's convnd function.)
This function takes an image and generates a matrix of all of its patches.
Given an array of filters, packs the filters into a DenseMatrix[Double] which has the following form: for a row i, column c+y*numChannels+x*numChannels*yDim corresponds to the pixel value at (x,y,c) in image i of the filters array.
Array of filters.
DenseMatrix of filters, as described above.