how to apply a convolution to 2D array

[jobo322]

May be the answer is obvious but I did not get the same result between ml-matrix-convolution and ml-convolution packages, so It need your help please.

I want to use this kernel

const smallFilter = [
  [0, 0, 1, 2, 2, 2, 1, 0, 0],
  [0, 1, 4, 7, 7, 7, 4, 1, 0],
  [1, 4, 5, 3, 0, 3, 5, 4, 1],
  [2, 7, 3, -12, -23, -12, 3, 7, 2],
  [2, 7, 0, -23, -40, -23, 0, 7, 2],
  [2, 7, 3, -12, -23, -12, 3, 7, 2],
  [1, 4, 5, 3, 0, 3, 5, 4, 1],
  [0, 1, 3, 7, 7, 7, 3, 1, 0],
  [0, 0, 1, 2, 2, 2, 1, 0, 0]
];

I understand that ml-convolution only accept 1D array so I flatten all. Is it the correct procedure?

note: would be great to be able to pass directly the 2D arrays for input data and kernel

[targos]

This package is for 1D convolution and doesn't support 2D arrays. It is possible to use it as an optimization for 2D if the kernel is separable, like we do here for images, but I'm not aware of a general way that could be implemented here directly.