Wavelet Shrinkage Multipatch Unbiased Distance Non-Local Adaptive Means

PROJECT TITLE :

Multipatch Unbiased Distance Non-Local Adaptive Means With Wavelet Shrinkage

ABSTRACT:

Many existing non-local means (NLM) approaches either utilise Euclidean distance to quantify the similarity between patches, or compute weight ij only once and maintain it unchanged over the future denoising repetitions, or use only the structure information of the denoised image to update weight ij. These may be the reasons for the poor denoising results. This study offers the non-local adaptive means (NLAM) for picture denoising in order to address these challenges. As an optimization variable, weight ij is continuously updated by NLAM. Introduced in the next section are three unbiased distances: Pixel, Patch and Coupling. These unbiased distances are more resilient than Euclidean distances in measuring the image pixel/patches' similarity. Our proposed non-local adaptive method is based on the unbiased distance (UD-NLAM). We offer multipatch UD-NLAM (MUD-NLAM) to accommodate varying noise levels because UD-NLAM employs just a single patch size to compute weight ij. We next propose a new denoising approach termed MUD-NLAM with wavelet shrinkage to significantly increase denoising performance (MUD-NLAM-WS). A comparison of the suggested NLAM, UD-NLAM, and MUD-NLAM with the current state-of-the-art denoising algorithms shows that MUDNLAM-WS outperforms them all.

Did you like this research project?

To get this research project Guidelines, Training and Code... Click Here