This paper presents an efficient denoising technique for removal of noise from digital images by combining filtering in both the transform (wavelet) domain and the spatial domain. The noise under consideration is AWGN and is treated as a Gaussian random variable. In this work the Karhunen-Loeve transform (PCA) is applied in wavelet packet domain that spreads the signal energy in to a few principal components, whereas noise is spread over all the transformed coefficients. This permits the

more »

... tion of a suitable shrinkage function on these new coefficients and elimination of noise without blurring the edges. The denoised image obtained by using the above algorithm is processed again in spatial domain by using total variation regularization. This post processing results in further improvement of the denoised results. Experimental results show better performance in terms of PSNR as compared to the performance of the methods when incorporated individually.