Implementation:Kornia Kornia Box Blur

Knowledge Sources	Kornia
Domains	Vision, Image_Filtering
Last Updated	2026-02-09 15:00 GMT

Overview

Applies box (mean) filtering to blur images using a uniform kernel where all weights are equal.

Description

This module is part of the Kornia library's filters subpackage. It provides the box_blur function and BoxBlur nn.Module class for smoothing images with a uniform averaging kernel. The box filter replaces each pixel with the average of its neighbors within the kernel window. The implementation supports both standard 2D convolution via filter2d and a more efficient separable decomposition via filter2d_separable, where the 2D box kernel is decomposed into two 1D passes.

Usage

Import this function or class when you need a simple, fast averaging blur for image smoothing, noise reduction, or as a building block in more complex filters such as guided filtering.

Code Reference

Source Location

• Repository: Kornia
• File: kornia/filters/blur.py
• Lines: 1-149

Signature

def box_blur(
    input: torch.Tensor,
    kernel_size: tuple[int, int] | int,
    border_type: str = "reflect",
    separable: bool = False,
) -> torch.Tensor: ...

class BoxBlur(nn.Module):
    def __init__(self, kernel_size: tuple[int, int] | int,
                 border_type: str = "reflect",
                 separable: bool = False) -> None: ...
    def forward(self, input: torch.Tensor) -> torch.Tensor: ...

Import

from kornia.filters import box_blur, BoxBlur

I/O Contract

Inputs

Name	Type	Required	Description
input	torch.Tensor (B, C, H, W)	Yes	The input image tensor to blur.
kernel_size	tuple[int, int] or int	Yes	The size of the box blurring kernel.
border_type	str	No (default "reflect")	Padding mode: constant, reflect, replicate, or circular.
separable	bool	No (default False)	If True, runs as a composition of two 1D convolutions for better performance.

Outputs

Name	Type	Description
output	torch.Tensor (B, C, H, W)	The blurred image, same shape as input.

Usage Examples

import torch
from kornia.filters import box_blur, BoxBlur

# Functional box blur
input = torch.rand(2, 4, 5, 7)
output = box_blur(input, (3, 3))
print(output.shape)  # torch.Size([2, 4, 5, 7])

# Separable box blur (more efficient for large kernels)
output = box_blur(input, (5, 5), separable=True)

# Module-based box blur
blur = BoxBlur((3, 3), border_type='reflect')
output = blur(input)
print(output.shape)  # torch.Size([2, 4, 5, 7])

Related Pages

• Principle:Kornia_Kornia_Mean_Filtering