Denoising of Image Using Least Minimum Mean Square Error
Vinay Dawar1, Mohit Bansal2
1Vinay Dawar, Electronics and Communication, D. R. College of Engineering and Technology, Sonepat, India.
3Mohit Bansal, Electronics and Communication, Bhagwan Mahaveer Institute of Engineering and Technology, Sonepat, India.
Manuscript received on September 27, 2012. | Revised Manuscript received on October 14, 2012. | Manuscript published on October 30, 2012. | PP: 69-73 | Volume-2 Issue-1, October 2012.  | Retrieval Number: A0727092112 /2012©BEIESP

Open Access | Ethics and Policies | Cite
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (

Abstract: In this paper, image denoising by linear minimum mean square-error estimation (LMMSE) scheme is proposed and also the determination of best suited wavelet for image denoising has been discussed. The over complete wavelet expansion (OWE) in noise reduction is used for taking the effective result instead of orthogonal wavelet transform. A vector has been designed by the combining the pixels at the same spatial location across scale to explore the strong inter-scale dependencies of OWE and apply LMMSE to the vector. Now, the performance evaluation of the proposed scheme is done by using different wavelet family. To measure the denoising performance, two criteria are used, first is signal information extraction and second is distribution error criterion. The best suite wavelet, which achieves best results between these two criteria, can be selected from wavelet family. To exploits the wavelet intrascale dependency and image discrimination, estimate the wavelet coefficients statistics and wavelet coefficient is classified by Context modelling.
Keywords: Linear Minimum Mean Square-Error estimation (LMMSE), over complete Wavelet Expansion (OWE).