Transformation based image de-noising

Abstract

In the processing of image de-noising, wavelet thresholding is an imported technique to refine the image from noisy components. Images are such as information which is transmitted from one source to another. During this transmission different noises are also added in the original information. The purpose is to introduce such type of technique to remove noises such as Gaussian noise from the images so that least extent of data carrying information diminish with the extreme removal of unwanted noisy components. We use Generalized Gaussian distribution for modeling. For the extraction of the original image, a hybrid scheme of thresholding is proposed. This algorithm is implemented and simulated in MATLAB using parameters like mean square error (MSE), peak signal to noise ratio (PSNR), visual quality and structural similarity index (SSIM). It is observed from the analysis that the hybrid technique gives result better than existing de-noising techniques. The computational complexity is less and provides better edge preserving during filtration.