An Approach for Image Denoising Based on Edge Soft-Judgement

This new approach employs different methods to estimate the edge and non-edge coefficients in the wavelet domain, separately. For the classification of the coefficients, the edge soft-judgement algorithm is introduced, which is according to both the gradient amplitude distribution and the calculation on the probability of the edge occurrence. When estimating the probability of the edge occurrence, a fast algorithm is educed. At last, the local-relation model is introduced to constrain the probability of the edge occurrence in scale-space to improve the estimation precision. Experimental results show that this new algorithm has a better performance and is easy to implement