spyrit.core.torch.walsh2_torch

spyrit.core.torch.walsh2_torch(img, H=None)[source]

Returns a 2D Walsh-ordered Hadamard transform of an image.

Args:

img (torch.tensor): Image to transform. Must have a shape of \((*, h, w)\) where \(h\) and \(w\) are the image height and width. The image must be square, i.e. \(h = w\). The image height and width must be a power of two.

H (torch.tensor, optional): 1D Walsh-ordered Hadamard transformation matrix. Specify this if you have already calculated the transformation matrix, or leave this field to None to use the Walsh-ordered fast Walsh-Hadamard transform. Defaults to None.

Returns:

torch.tensor: Hadamard-transformed image. Has same shape as input image.

..warning::

This function is deprecated and has been moved to spyrit.core.torch. It is recommended to use spyrit.core.torch.fwht_2d() instead for natural- or walsh-ordered Hadamard transforms.

Example:

>>> img = torch.rand(1, 3, 64, 64) # (batch, channels, height, width)
>>> img_transformed = walsh2_torch(img)
>>> img_transformed.shape
torch.Size([1, 3, 64, 64])