*2.2. Non-Subsampled Shearlet Transform*

Because Shearlet transform does not have the translational invariance, it is easy to generate the pseudo-Gibbs phenomenon near singular points when it is applied to image fusing and denoising. The NSST with the translational invariance can solve this problem. The discretization process of the NSST is composed of the multi-scale subdivision and the direction localization:


**Figure 2.** The decomposition process of the NSST.
