A New Wavelet-based Image Denoising Using Undecimated Discrete Wavelet Transform and Least Squares Support Vector Machine
Xiang-Yang Wang,Hong-Ying Yang,Zhong-Kai Fu
DOI: https://doi.org/10.1016/j.eswa.2010.03.014
IF: 8.5
2010-01-01
Expert Systems with Applications
Abstract:Image denoising is an important image processing task, both as itself, and as a preprocessing in image processing pipeline. The least squares support vector machine (LS-SVM) has shown to exhibit excellent classification performance in many applications. Based on undecimated discrete wavelet transform, a new wavelet-based image denoising using LS-SVM is proposed in this paper. Firstly, the noisy image is decomposed into different subbands of frequency and orientation responses using the undecimated discrete wavelet transform. Secondly, the feature vector for a pixel in a noisy image is formed by the spatial regularity in wavelet domain, and the LS-SVM model is obtained by training. Then the wavelet coefficients are divided into two classes (noisy coefficients and noise-free ones) by LS-SVM training model. Finally, all noisy wavelet coefficients are relatively well denoised by shrink method, in which the adaptive threshold is utilized. Extensive experimental results demonstrate that our method can obtain better performances in terms of both subjective and objective evaluations than those state-of-the-art denoising techniques. Especially, the proposed method can preserve edges very well while removing noise.