Comparative analysis between non-linear wavelet based image denoising techniques

Authors

  • Milan Chikanbanjar Department of Computer Engineering, Khwopa Engineering College, Libali-8, Bhaktapur, Nepal

DOI:

https://doi.org/10.3126/jsce.v5i0.22373

Keywords:

Image denoising, Gaussian noise, salt and pepper noise, discrete wavelet transform, thresholding, non-linear wavelet techniquesw, Peak-Signal-to-Noise-Ratio (PSNR), SSIM (Structural Similarity Index)

Abstract

Digital images have been a major form of transmission of visual information, but due to the presence of noise, the image gets corrupted. Thus, processing of the received image needs to be done before being used in an application. Denoising of image involves data manipulation to remove noise in order to produce a good quality image retaining different details. Quantitative measures have been used to show the improvement in the quality of the restored image by the use of various thresholding techniques by the use of parameters mainly, MSE (Mean Square Error), PSNR (Peak-Signal-to-Noise-Ratio) and SSIM (Structural Similarity index). Here, non-linear wavelet transform denoising techniques of natural images are studied, analyzed and compared using thresholding techniques such as soft, hard, semi-soft, LevelShrink, SUREShrink, VisuShrink and BayesShrink. On most of the tests, PSNR and SSIM values for LevelShrink Hard thresholding method is higher as compared to other thresholding methods. For instance, from tests PSNR and SSIM values of lena image for VISUShrink Hard, VISUShrink Soft, VISUShrink Semi Soft, LevelShrink Hard, LevelShrink Soft, LevelShrink Semi Soft, SUREShrink, BayesShrink thresholding methods at the variance of 10 are 23.82, 16.51, 23.25, 24.48, 23.25, 20.67, 23.42, 23.14 and 0.28, 0.28, 0.28, 0.29, 0.22, 0.25, 0.16 respectively which shows that the PSNR and SSIM values for LevelShrink Hard thresholding method is higher as compared to other thresholding methods, and so on. Thus, it can be stated that the performance of LevelShrink Hard thresholding method is better on most of tests.

Downloads

Download data is not yet available.
Abstract
1356
PDF
315

Downloads

Published

2018-08-31

How to Cite

Chikanbanjar, M. (2018). Comparative analysis between non-linear wavelet based image denoising techniques. Journal of Science and Engineering, 5, 58–67. https://doi.org/10.3126/jsce.v5i0.22373

Issue

Section

Research Papers