Abstract: The process of reconstructing the original image which is corrupted by noise is called de-noising. Discrete wavelet transform is superior over other transforms in terms of image denoising as its functions are localized both in frequency and time domain. The denoising of image has three major steps decomposition, thresholding and reconstruction. In this study, the qualitative and quantitative assessment of the denoised image is done on the basis of different noise variance and further analytic result shows that soft threshold is superior over hard threshold in terms of matrices such as Root Mean Square Error (RMSE) and Peak Signal-to-Noise Ratio (PSNR).