High dimension signal denoising using wavelet transform

Abstract

Image denoising is one of the classical problems in digital image processing and has been studied for nearly half a century due to its important role as a pre-processing step in various electronic imaging applications. The search for efficient image denoising methods is still a valid challenge. In spite of the sophistication of the recently proposed methods, most algorithms have not yet attained a desirable level of applicability. The purpose is to introduce such type of technique to remove Gaussian noise from the image so that least extent of data carrying information diminishes with removal of unwanted noisy components. An improved adaptive wavelet threshold function is designed for the extraction of original image. The wavelet decomposing detail coefficients of the image mixed with Gaussian noise are denoised by this improved threshold function, then reconstructed together with the wavelet decomposing approximation coefficients to get the denoised image. The experimental results show that the denoising effect of the improved threshold function is superior to hard threshold and soft threshold. Different thresholds will be set for wavelet details of each level when denoising. Considering the advantage of modified bilateral filter, this thesis proposes to combine wavelet improved threshold denoising with modified bilateral filtering, termed as combined denoising method. MATLAB simulation results show that, this method has better effect on removing Gaussian noise mixed in images. The results have compared using peak signal to noise ratio (PSNR), visual quality, structural similarity index (SSIM) parameters. Better denoising effect demonstrates the adaptability of the combined denoising method.