Compressed sensing by inverse scale space and curvelet thresholding

Compressed sensing provides a new sampling theory for data acquisition, which says that compressible signals can be exactly reconstructed from highly incomplete sets of linear measurements. It is significant to many applications, e.g., medical imaging and remote sensing, especially for measurements limited by physical and physiological constraints, or extremely expensive. In this paper, we proposed a recovery algorithm from a view of reaction-diffusion equations, by applying curvelet thresholding in inverse scale space flows. Numerical experiments in medical CT and aerospace remote sensing show its good performances for recovery of detailed features from incomplete and inaccurate measurements, in comparison with some existing methods.