Abstract—Image denoising methods based on Markov
random field (MRF) are often shown over-smooth phenomenon
for strong noise image. Wavelet analysis has good
time-frequency local ability and preserves the image edge
information well for image denoising problem. Based on
wavelet analysis and MRF theory, we propose a wavelet markov
field of experts (WMFoE) framework to deal with image
denoising problems. The noise image is divided into
low-frequency and high-frequency component, and MRF are
used to deal with low-frequency component. For
high-frequency component, a clustering based soft-threshold
method is used to remove the noise signal. Then, the restored
image can be gotten by reconstruction from different
components. Experiment results show that our method not only
gets good PSNR and SSIM values but also preserves image edge
information especially for strong noise image, compared with
BM3D etc. state-of-art methods.
Index Terms—Markov random field, image denoising, wavelet transformation, k-means clustering.
Xiang Wu, Zhiguo Shi, and Lei Liu with the School of Computer and Communication Engineering, University of Science and Technology Beijing, China (e-mail: firstname.lastname@example.org, email@example.com, firstname.lastname@example.org).
Cite: Xiang Wu, Zhi-Guo Shi, and Lei Liu, "Image Denoising with Wavelet Markov Fields of Experts," International Journal of Machine Learning and Computing vol. 4, no. 6, pp. 527-532, 2014.