Image denoising in wavelet domain with filtering and thresholding

Abstract

In this paper, the proposed method is implemented for removal of salt & pepper and Gaussian noise of black & white & color images to acquire the quality output. In this work initially wavelet coefficients are extracted for noisy images. Later apply denoise filtering technique on the high transform sub bands of noisy images (either color/ B & W) using new laplacian filters with 4 directions. Finally threshold of an image is generated to extract denoisy coefficients. At last inverse of above subband coefficients can give denoise image for further processing. The proposed method is verified against various B & W/color images and it gives a better PSNR (Peak Signal to Noise Ratio) & MI (Mutual Information). These values are compared with different noise densities and analyzed visually.