Sparsity assisted phase retrieval of complex valued objects

Abstract

Iterative phase retrieval of complex valued objects (phase objects) suffers from twin image problem due to the presence of features of image and its complex conjugate in the recovered solution. The twin-image problem becomes more severe when object support is centro-symmetric. In this paper, we demonstrate that by modifying standard Hybrid-Input output (HIO) algorithm using an adaptive sparsity enhancement step, the twin-image problem can be addressed successfully even when the object support is centro-symmetric. Adaptive sparsity enhanced algorithm and numerical simulation for binary as well as gray scale phase objects are presented. The high quality phase recovery results presented here show the effectiveness of adaptive sparsity enhanced algorithm.

Experiments on sparsity assisted phase retrieval of phase objects

“…In our experiments we observed that in the first 10 or so initial outer iterations, the value of N was approximately in the range 20-30 which later reduced to approximately 5 as iterations progressed. We have already shown the effectiveness of this approach in a simulation study [23]. The adaptive approach as used here has appeared in the image reconstruction literature before [24] and is effective in balancing data consistency (that is imposed by GS or OSS-HIO) with sparsity penalty without introducing additional free parameters in the algorithm.…”