Research of Image Denoising Based on Edge Preservation

In this paper, a new algorithm for removing noise from digital images, based on the Stationary Wavelet Transform and Mathematic Morphology, is proposed. The level-dependent thresholding scheme is used to extract the multiscale edge. And apply the dilation operation to expand the edge location information, for the sake of preserving the important wavelet coefficient and its neighborhood coefficients. This process will improve the preservation of edges significantly while maintaining the smoothness on non-edge regions. It demonstrated through simulations with images contaminated by additive white Gaussian noise that this algorithm produced acceptable results, both visually and in terms of signal-to-noise-ratio.