Low-complexity, hierarchically-adapted wavelet thresholding for image denoising

We present a low-complexity wavelet thresholding scheme that utilizes the correlation that exists between coefficients belonging to subsequent resolution levels in a multi-scale representation of images. The proposed threshold adapts to the data and noise statistics in each subband. Spatial adaptation for each subband coefficient is achieved by making use of its corresponding parent coefficient. Experimental results show that the proposed hierarchically-adapted threshold outperforms several popular thresholding schemes for image denoising, both in terms of PSNR and visual quality.