Abstract: To improve the performance of the noise variance estimation, an algorithm based on the wavelet coefficients estimation of original image are proposed. Using the proposed approach, the authors exploited the inter-correlation of adjacent wavelet scales to estimate the wavelet coefficients of original image, thus got the purer noise coefficients by deducting the estimated values from the coefficients of the noisy image, then estimated once again using the Donoho's formula. Results show that the proposed approach outperforms traditional ones, especially in scenarios where there is less noise or much more image details.