Tomographic reconstruction based on phase retrieval and compressive sensing

Abstract

Tomographic reconstruction is ill-posed for coherent reconstruction methods due to the higher dimensionality of the 3-D complex wavefield recorded using a single 2-D intensity measurement. This limitation is partially addressed in this paper by incorporating compressive sensing in conjunction with the multiple-plane phase retrieval reconstruction method. Simulations were performed to demonstrate the effectiveness of the technique in isolating object points at predetermined axial positions. The interplane crosstalk distortions are minimized using the compressive sensing framework resulting to enhanced reconstructions with more apparent separation of object points.