Fast Noise Level Estimation Algorithm Based on Two-Stage Support Vector Regression

Although the noise level estimation algorithms based on principal component analysis are of high estimation accuracy,their iterative process that selects the homogeneous patches from the raw patches leads to low efficiency.To improve the efficiency of the existing algorithms,we proposed a new fast NLE algorithm based on two-stage support vector regression.Considering the fact that the first several eigenvalues extracted from the covariance matrix of raw patches are significantly correlated with the noise level,we first estimated the noise level roughly with a coarse prediction model trained with support vector machine technique;then,according to the preliminary estimation result,the final noise level was obtained with a more accurate prediction model that is specifically trained for low,medium,or high noise levels.Extensive experiments show that,the proposed algorithm greatly improves the execution efficiency without reducing the estimation accuracy too much.Compared with other state-of-the-art algorithms,the proposed algorithm has the obvious advantages in terms of both accuracy and efficiency,when it is used as the preprocessing module of various image processing algorithms.