Energy Pixel Rule for Image Denoising Based on Slantlet Transform and K-SVD Algorithm

Rasheed Humood, Waleed

	Energy Pixel Rule for Image Denoising Based on Slantlet Transform and K-SVD Algorithm
Iraqi Journal of Information Technology
Article 1, Volume 6, Issue 3, 2014, Pages 50-57
Author
Waleed Rasheed Humood
Abstract
All digital images contain some degree of noise. Image denoising algorithm attempts to remove this noise from the image. Ideally, the resulting de-noised image will not contain any noise or added artifacts. Denoising of natural images corrupted by Gaussian noise using Slantlet transformation techniques is very effective because of its ability to capture the energy of a signal in few energy transform values. In this paper,mean pixel energy for each windows of noisy image will calculate by using the Energy Estimation operation then store in a dictionary. Noisy image decomposed using Slantlet transform and then applied K-SVD algorithm noise removal. In the same time, the modified mean filter is also applied to the de-noised imagefor noise removal.Finally, the de-noised image get from the comparison between (modified mean with K-SVD algorithm) outputs and select the appropriate pixels depend on the energy value stored on the dictionary. From the experimentalresults, the proposed method can efficiently remove a single Gaussian noise while preserving the image information well with less blur and demonstrate its better fidelity criteria (PSNR and ERMS) after apply proposed pixel rule selection compared with results of modified mean and K-SVD algorithm.
Keywords
image; Denoising; Slantlet; Image denoising; SVD

Statistics Article View: 196 PDF Download: 35