Wavelets Approach in Choosing Adaptive Regularization Parameter

In noise removal by the approach of regularization, the regularization parameter is global. Constructing the variational model min g ?f - g?L2(R)2 +?R(g), g is in some wavelets space. Through the wavelets pyramidal decompose and the different time-frequency properties between noise and signal, the regularization parameter is adaptively chosen, the different parameter is chosen in different level for adaptively noise removal.