Estimating Noise Parameter Based on the Wavelet Coefficients Estimation of Original Image

On the basis of Donoho´s traditional method, a novel approach is proposed to estimate variance of Gaussian white noise based on the wavelet coefficients estimation of the original image. The approach makes use of the wavelet coefficients estimation of the original image to obtain more pure noise coefficients, then does estimation once again based on the Donoho´s formula. Simulation results are presented which indicate that the proposed approach is better than traditional ones, especially in scenarios where there is less noise or much more image details.