Convolve2d just by using Numpy

Question

I am studying image-processing using Numpy and facing a problem with filtering with convolution.

I would like to convolve a gray-scale image. (convolve a 2d A

Answer 1

Cleaned up using as_strided and @Crispin 's einsum trick from above. Enforces the filter size into the expanded shape. Should even allow non-square inputs if the indices are compatible.

def conv2d(a, f):
    s = f.shape + tuple(np.subtract(a.shape, f.shape) + 1)
    strd = numpy.lib.stride_tricks.as_strided
    subM = strd(a, shape = s, strides = a.strides * 2)
    return np.einsum('ij,ijkl->kl', f, subM)