Synchrosqueezed curvelet transform for two-dimensional mode decomposition (Q2878992)

In some applications, a 2-dimensional signal is a superposition of several banded wave-like components, and it is required to separate these components. In this paper, the authors introduce the so-called synchrosqueezed curvelet transform (SSCT) as an optimal tool to estimate local, well-separated wave-vectors of banded wave-like components. The SSCT consists of a generalized curvelet transform with geometric scaling parameters, and a synchrosqueezing technique for a sharpened phase space representation. A discrete analogue of SSCT and a corresponding fast algorithm are presented too. Numerical examples with synthetic and real data demonstrate the performance of SSCT.