from __future__ import annotations
from kornia.core import Module, Tensor
from kornia.core.check import KORNIA_CHECK_IS_TENSOR
from .filter import filter2d, filter2d_separable
from .kernels import _unpack_2d_ks, get_box_kernel1d, get_box_kernel2d
[docs]def box_blur(
input: Tensor, kernel_size: tuple[int, int] | int, border_type: str = 'reflect', separable: bool = False
) -> Tensor:
r"""Blur an image using the box filter.
.. image:: _static/img/box_blur.png
The function smooths an image using the kernel:
.. math::
K = \frac{1}{\text{kernel_size}_x * \text{kernel_size}_y}
\begin{bmatrix}
1 & 1 & 1 & \cdots & 1 & 1 \\
1 & 1 & 1 & \cdots & 1 & 1 \\
\vdots & \vdots & \vdots & \ddots & \vdots & \vdots \\
1 & 1 & 1 & \cdots & 1 & 1 \\
\end{bmatrix}
Args:
image: the image to blur with shape :math:`(B,C,H,W)`.
kernel_size: the blurring kernel size.
border_type: the padding mode to be applied before convolving.
The expected modes are: ``'constant'``, ``'reflect'``, ``'replicate'`` or ``'circular'``.
separable: run as composition of two 1d-convolutions.
Returns:
the blurred tensor with shape :math:`(B,C,H,W)`.
.. note::
See a working example `here <https://kornia.github.io/tutorials/nbs/filtering_operators.html>`__.
Example:
>>> input = torch.rand(2, 4, 5, 7)
>>> output = box_blur(input, (3, 3)) # 2x4x5x7
>>> output.shape
torch.Size([2, 4, 5, 7])
"""
KORNIA_CHECK_IS_TENSOR(input)
if separable:
ky, kx = _unpack_2d_ks(kernel_size)
kernel_y = get_box_kernel1d(ky, device=input.device, dtype=input.dtype)
kernel_x = get_box_kernel1d(kx, device=input.device, dtype=input.dtype)
out = filter2d_separable(input, kernel_x, kernel_y, border_type)
else:
kernel = get_box_kernel2d(kernel_size, device=input.device, dtype=input.dtype)
out = filter2d(input, kernel, border_type)
return out
[docs]class BoxBlur(Module):
r"""Blur an image using the box filter.
The function smooths an image using the kernel:
.. math::
K = \frac{1}{\text{kernel_size}_x * \text{kernel_size}_y}
\begin{bmatrix}
1 & 1 & 1 & \cdots & 1 & 1 \\
1 & 1 & 1 & \cdots & 1 & 1 \\
\vdots & \vdots & \vdots & \ddots & \vdots & \vdots \\
1 & 1 & 1 & \cdots & 1 & 1 \\
\end{bmatrix}
Args:
kernel_size: the blurring kernel size.
border_type: the padding mode to be applied before convolving.
The expected modes are: ``'constant'``, ``'reflect'``,
``'replicate'`` or ``'circular'``. Default: ``'reflect'``.
separable: run as composition of two 1d-convolutions.
Returns:
the blurred input tensor.
Shape:
- Input: :math:`(B, C, H, W)`
- Output: :math:`(B, C, H, W)`
Example:
>>> input = torch.rand(2, 4, 5, 7)
>>> blur = BoxBlur((3, 3))
>>> output = blur(input) # 2x4x5x7
>>> output.shape
torch.Size([2, 4, 5, 7])
"""
def __init__(
self, kernel_size: tuple[int, int] | int, border_type: str = 'reflect', separable: bool = False
) -> None:
super().__init__()
self.kernel_size = kernel_size
self.border_type = border_type
self.separable = separable
if separable:
ky, kx = _unpack_2d_ks(self.kernel_size)
self.register_buffer("kernel_y", get_box_kernel1d(ky))
self.register_buffer("kernel_x", get_box_kernel1d(kx))
self.kernel_y: Tensor
self.kernel_x: Tensor
else:
self.register_buffer("kernel", get_box_kernel2d(kernel_size))
self.kernel: Tensor
def __repr__(self) -> str:
return (
f"{self.__class__.__name__}"
f"(kernel_size={self.kernel_size}, "
f"border_type={self.border_type}, "
f"separable={self.separable})"
)
def forward(self, input: Tensor) -> Tensor:
KORNIA_CHECK_IS_TENSOR(input)
if self.separable:
return filter2d_separable(input, self.kernel_x, self.kernel_y, self.border_type)
return filter2d(input, self.kernel, self.border_type)