An Efficient Adaptive ROAD And Robust Estimation Based De-Noising Technique

In this paper, an algorithm for de-noising fixed value impulse noise is presented. The ROAD(Rank Ordered Absolute Difference) value computed using fixed window size 3X3 detects impulse pixel. The corrupted pixels are identified and replaced by the estimated value by Lorentzian estimator. The window size for estimation is adaptive and expands up to 7 X 7 in case of high noise densities. The noise free pixels remain unchanged. The performance of the algorithm is evaluated using PSNR and MSE. The results prove that the method works well for high density impulse noise, preserves edges and other fine details in the image.