Medical and Natural Image Denoising by Split Bregman’s Total Variation Filtering with Wavelet Thresholding

In this paper a new method of image denoising is proposed. This method combines two very popular techniques of denoising, Wavelet decomposition and total variation filtering. We will use this method for denoising of medical images as well as natural images. In this algorithm, a noisy image is first decomposed in its wavelet coefficients. Then these coefficients filtered by soft thresholding. This threshold is calculated by total variation method. Inverse wavelet transform of filtered and modified coefficients of image give the reconstruction of denoised image. Total variation minimization filter will sufficiently remove noisy coefficients and retain fine edge information of image. Very few iteration of TV will produce denoised image with edge information. Split bregman method is applied for TV minimization. In terms of PSNR values, this combination will produces better results.