Cognitive Behavior and Human Computer Interaction Based on Machine Learning Algorithms

4Denoising of Digital Images Using Wavelet-Based Thresholding Techniques: A Comparison

Devanand Bhonsle

SSTC, SSGI, Bhilai, India

Email: [e-mail]

Abstract

Noise is an unwanted entity that corrupts the image due to which its quality degrades. This degradation may be suppressing details like edges, ridges, and other fine structures that are important to analyze the details. Hence, removal of noise is essential or at least its effect must be reduced to a certain extent. The process by which an image reduces the noise is called “Image Denoising”. It is a filtration technique that suppresses the noise and restores the image. Gaussian noise is one of the noise signals, which corrupts the signals almost in all the frequency band; hence, it is known as white Gaussian noise. Since it is additive, i.e., it is added to all the image’s pixel values, Additive White Gaussian Noise. In this chapter, Gaussian noise suppression using various wavelet thresholding has been discussed. Results exhibit the capability of wavelet-based filters to reduce the effect of Gaussian noise to an acceptable extent.

Keywords: Noise, wavelet transform, peak signal to noise ratio, thresholding,

mean square error, structural similarity index matrix

4.1 Introduction

Noise may be introduced due to various reasons. Mainly, noise introduces in the images during the acquisition and transmission from source to destination [1]. Environmental condition and quality of sensing elements are the major ...