Adaptive image denoising using wavelet thresholding

In order to preserve fine details in image denoising, we propose a scheme by assuming that the deviations of the noisy and the original wavelet coefficients of image are not always the same across the scales. The proposed algorithm considers not only the correlation of inter-scale wavelet coefficients but also the mentioned assumptions. In the process of denoising, the proposed denoising threshold can adaptively adjust itself on the basis of its position and decomposition scale. We demonstrate its effectiveness through simulations with images contaminated by additive white Gaussian noise and compare it with the classical threshold method. Experimental results show that the performance of our method can preserve image details well both in visual effect and in terms of peak signal-to-noise ratio.