Sequence to Sequence Similarity for Image Denoising

Author(s):

Akhila A S , MARIAN COLLEGE OF ENGINEERING; Dr. M Sasikumar, MARIAN COLLEGE OF ENGINEERING

Keywords:

Image denoising, block similarity, sequence to sequence similarity, edge information

Abstract:

Image denoising is a continuing field of researches. Despite of several remarkable progresses in image denoising technologies, still there exists so many challenges. None of the denoising algorithm could be widely used for all applications. Fundamental challenge in a denoising technique is that it should adequately smooth a noisy image to efficiently remove the noise, while simultaneously preserving the important structural details within the image. In this work, we present a new similarity concept for image denoising; sequence to sequence similarity. In this concept of similarity, instead of taking pixel averages and block similarities, which are traditional concepts, sequence to sequence similarity is evaluated. This measure not only denoises the image but also improves the image quality by preserving the structural contents in an image. The experimental results show that the proposed method is good in preserving the edge information comparing to other approaches using pixel to pixel similarity and block to block similarity.

Other Details:

Manuscript Id	:	IJSTEV3I9091
Published in	:	Volume : 3, Issue : 9
Publication Date	:	01/04/2017
Page(s)	:	201-206

Download Article