A hybrid algorithm using transform for de-noising celestial images

Image de-noising is the process of minimizing blurring effects in an image which might possibly occur due to atmospheric turbulence, motion-al disturbance and out of focus camera. Image de-noising is needed to obtain a clearer image and make it more useful as an information source. Few of the commonly added noise include salt and pepper, Gaussian, speckle noise. An efficient technique de-noising is filtering performed either in the spatial or frequency domain. Among the non-linear filters, Adaptive Weighted Median Filter (A WMF) reduces noise, preserving few of the image details, however, they do not considerably preserve edges. Frequency domain filtering using transforms as the Discrete Fourier Transform preserves edges, but does not provide any temporal details. This paced up the wavelets to be used for de-noising purpose. The Discrete form of Wavelet Transform (DWT) help represent an image with fewer coefficients, thereby eliminating redundancy. Using the DWT provides good results with approximation and details in all dimensions. A Hybrid algorithm is presented in this paper, in which various salient features of the DWT and AWMF methods are being considered. The proposed algorithm works by primarily applying the DWT on a corrupt noisy image with predefined levels of extraction and thresholding and then the resulting image is further processed using the AWMF for better quality. The proposed method provides sufficient image details and is computationally efficient taking up less memory as compared to AWMF and DWT. The quality of images de-noised by the use of the Hybrid Algorithm is studied and presented in terms of parameters such as the peak signal to noise ratio (PSNR), similarity index (SSIM) and the mean squared error (MSE).