Image Denoising by Multiscale – LMMSE in Wavelet Domain and Joint Bilateral Filter in Spatial Domain
M.Vijay1, L.Saranya Devi2
1M. .Vijay, Department of ECE, PSRR College of Engg, Sivakasi, India.
2L.Saranya Devi, Department of ECE, PSRR College of Engg, Sivakasi, India.
Manuscript received on February 15, 2012. | Revised Manuscript received on February 20, 2012. | Manuscript published on March 05, 2012. | PP: 411-416 | Volume-2 Issue-1, March 2012. | Retrieval Number: A0467022112/2012©BEIESP
Open Access | Ethics and Policies | Cite
© The Authors. Published By: Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: This paper deals with LMMSE-based denoising scheme with a wavelet interscale model and Joint bilateral Filter in spatial domain. The proposed algorithm consists of two stages .In the first stage, a vector is represented by the wavelet coefficients at the same spatial locations at two adjacent scales and the LMMSE is applied to the vector. Compare to Orthogonal Wavelet Transform (OWT), Overcomplete Wavelet Expansion (OWE) provides better results hence it is employed. While applying the LMMSE rule, the important features in an image like edges, curves and textures can be identified. Also spatial domain method output provides a high quality denoising image than wavelet method with fewer artifacts; hence this wavelet domain output as a reference image for the Joint Bilateral Filter (JBF) .By using this reference image and the non-linear combination of information of adjacent pixel, the edge details of the images can be preserved in a well manner. The experimental results prove that the proposed approach is competitive when compared to other denoising methods in reducing various types of noise. Also the proposed algorithm outperforms other methods both visually and in case of objective quality peak-signal-to-noise ratio (PSNR).
Keywords: Image Denoising; Joint Bilateral Filter; Multiscale LMMSE; Interscale Wavelet Model.