# WienerPytorch

**Repository Path**: uho163/WienerPytorch

## Basic Information

- **Project Name**: WienerPytorch
- **Description**: No description available
- **Primary Language**: Unknown
- **License**: MIT
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2025-10-29
- **Last Updated**: 2025-10-29

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# WienerPytorch
Pytorch implementation of the Wiener image denoising filter. Allow models to be trained through the filter and benefit from CUDA speedup. 

Uses 3D FFT, custom Gaussian windowing,  tunable overlapping block density. 

## Usage
In `test_denoiser.py` replace `dataset_path` with location of noisy images directories. As dataloader is designed for multiple image folders, your directory should follow the structure below: 
```
Base
│
└───dataset_path
│   │
│   └───subfolder1
│       │   file111.png
│       │   file112.png
│   │   
│   └───subfolder2
│       │   file111.png
│       │   file112.png
```

`std` specifies the noisy image standard deviation.

`win_size` specifies the block size for the Wiener filter.

`overlap` specifies the number of overlapping windows in a given block. In other words, selects the sliding window stride. E.g. `win_size = 32` and `overlap = 4`  creates a stride of 8 pixels for the windows.

Denoised images saved to current directory.

## Citation

If you use this repository in your research or software, **please cite it** and acknowledge the author.

**Software**

> Bled, Clément. *WienerPytorch: PyTorch Wiener Denoising Filter*. GitHub. https://github.com/MrBled/WienerPytorch

You can also use the BibTeX from the **Cite this repository** panel on the right (powered by `CITATION.cff`).

**Related paper**

> Clément Bled and François Pitié. “Pushing the Limits of the Wiener Filter in Image Denoising.” *2023 IEEE International Conference on Image Processing (ICIP)*, 2023. https://doi.org/10.1109/ICIP49359.2023.10222826


## Denoised Gaussian Noise Images, STD=50 
![
](images/13_noisy.png)

![
](images/13_denoise.png)

![
](images/5_noisy.png)

![
](images/5_denoise.png)

![
](images/34_noisy.png)

![
](images/34_denoise.png)