denoising algorithm by using Curvelet Transform:
1. Compute all thresholds for cuvelets;
 2. Compute norm of curvelets;
3. Apply curvelet transform to noisy image;
4. Apply hard thresholding to the curvelet coefficients; and
5. Apply inverse cuvrelet transform to the result of step 4.

I do not understand step 2. In your opinion, Is not more appropriate 
calculate norm of each curvelet coefficient after the transform image to 
curvelet. If there is a scientific reason for step 2, please explain. 
Maybe I just did not notice