Non-Local Means

Pre-Processing > Non-Local Means

Reduces image noise using an advanced filtering algorithm [1]. Each pixel’‘s local window is compared to windows around it. The windows’ center pixels are averaged together with weights depending on the variations between the windows.

1. Window Size

Size of neighborhood (in pixels) which is considered about each pixel for the filter (RECOMMENDED: 2-6)

2. Strength

Factor which controls the strength of the filter (Recommended: 0.2-0.6)

References

[1] Buades, A., B. Coll, and J.-M. Morel. “A Non-Local Algorithm for Image Denoising.” 2005 IEEE® Computer Society Conference on Computer Vision and Pattern Recognition. Vol. 2, June 2005, pp. 60–65.

Tips

• Noise-reduction filters are often very useful for removing granular noise while maintaining clean feature boundaries.
• We recommend you try all of the filters to find which once works best.
• Non-Local Means is one of the more powerful pre-processing filters MIPAR has to offer, but can take longer to execute for larger images.
• Starting Window Size Value: 2-4, Strength: 0.2, decrease Strength for low resolution images. Increasing window size above 20 increases time of execution with little changes.