DETAIL PRESERVED DENOISING ALGORITHM VIA PATCH BASED EDGE SIMILARITY INDEX AND JOINT BILATERAL FILTER
Abstract
vioft2nntf2t|tblJournal|Abstract_paper|0xf4ff1e362c000000af7e0b0001000900
Denoising algorithms are getting more attention in these days because they play a vital role in other applications of image processing. As details structures are the important information of Human Visual System, its preservation in denoised image is highly demanding one. A two step denoising algorithm based on Patch based Edge Similarity Index and Joint Bilateral Filter algorithm proposed in this paper, preserves the edge structures and produce visually pleasant denoised image. Edge Similarity Index (ESI) proposed in the paper group patches according to their similarity in orientation and hence preserve the detail feature. The optimally grouped patches transferred to Principal Component Analysis (PCA) domain and the proposed noise suppression method eliminates the noisy component. Adaptive soft thresholding noise suppression method suppress the noise based on local noise estimation. Noise estimation in local level helps to estimate the noise accurately where noise affected differently in regions of a scene. Strong noises residuals may exist after first step, the denoised image in the first step further processed by a Joint Bilateral Filter for producing visually pleasant denoised image. Experimental results shown that proposed denoising algorithm achieves comparable detail preserving performance in terms of visual analysis and quantitative analysis over other state of art method.

Authors
Dayana David
Kerala Agricultural University, India

Keywords
Fixed Patch, Variable Patch, Denoising Patch, Edge Similarity Index, Joint Bilateral Filter
Yearly Full Views
JanuaryFebruaryMarchAprilMayJuneJulyAugustSeptemberOctoberNovemberDecember
110000000000
Published By :
ICTACT
Published In :
ICTACT Journal on Image and Video Processing
( Volume: 12 , Issue: 3 , Pages: 2661-2668 )
Date of Publication :
February 2022
Page Views :
202
Full Text Views :
3

Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.