from __future__ import annotations
from kornia.core import Module, Tensor
from .gaussian import gaussian_blur2d
[docs]def unsharp_mask(
input: Tensor, kernel_size: tuple[int, int] | int, sigma: tuple[float, float] | Tensor, border_type: str = 'reflect'
) -> Tensor:
r"""Create an operator that sharpens a tensor by applying operation out = 2 * image - gaussian_blur2d(image).
.. image:: _static/img/unsharp_mask.png
Args:
input: the input tensor with shape :math:`(B,C,H,W)`.
kernel_size: the size of the kernel.
sigma: the standard deviation of the kernel.
border_type: the padding mode to be applied before convolving.
The expected modes are: ``'constant'``, ``'reflect'``,
``'replicate'`` or ``'circular'``.
Returns:
the blurred tensor with shape :math:`(B,C,H,W)`.
Examples:
>>> input = torch.rand(2, 4, 5, 5)
>>> output = unsharp_mask(input, (3, 3), (1.5, 1.5))
>>> output.shape
torch.Size([2, 4, 5, 5])
"""
data_blur: Tensor = gaussian_blur2d(input, kernel_size, sigma, border_type)
data_sharpened: Tensor = input + (input - data_blur)
return data_sharpened
[docs]class UnsharpMask(Module):
r"""Create an operator that sharpens image with: out = 2 * image - gaussian_blur2d(image).
Args:
kernel_size: the size of the kernel.
sigma: the standard deviation of the kernel.
border_type: the padding mode to be applied before convolving.
The expected modes are: ``'constant'``, ``'reflect'``,
``'replicate'`` or ``'circular'``.
Returns:
the sharpened tensor with shape :math:`(B,C,H,W)`.
Shape:
- Input: :math:`(B, C, H, W)`
- Output: :math:`(B, C, H, W)`
.. note::
See a working example `here <https://kornia.github.io/tutorials/nbs/unsharp_mask.html>`__.
Examples:
>>> input = torch.rand(2, 4, 5, 5)
>>> sharpen = UnsharpMask((3, 3), (1.5, 1.5))
>>> output = sharpen(input)
>>> output.shape
torch.Size([2, 4, 5, 5])
"""
def __init__(
self, kernel_size: tuple[int, int] | int, sigma: tuple[float, float] | Tensor, border_type: str = 'reflect'
) -> None:
super().__init__()
self.kernel_size = kernel_size
self.sigma = sigma
self.border_type = border_type
def forward(self, input: Tensor) -> Tensor:
return unsharp_mask(input, self.kernel_size, self.sigma, self.border_type)