Illumination schemes for coded coherent diffraction imaging: A comprehensive comparison

Coded coherent diffraction imaging (CDI) introduces physical constraints to the conventional CDI by wavefront modulation, providing data redundancy for successful phase retrieval. In coded CDI, an object is modulated by several pre-defined illumination patterns that are either real or complex. The iterative phase retrieval algorithms are typically adopted to reconstruct the object wavefront. We note that different illumination schemes might affect imaging efficiency and achievable reconstruction quality. However, there has been no comprehensive review discussing respective advantages, which is important for further applications and development of coded CDI. In this study, we comprehensively investigate six representative illumination pattern structures in a unified framework. The illumination patterns adopted for both amplitude and phase modulation include binarized random, binarized blue noise, complementary, incremental, gray-scale random, and gray-scale blue noise patterns. To provide a fair platform from an algorithmic point of view, the reconstruction algorithms are based on the ptychographic iterative engine (PIE) framework with both serial and parallel structures. The comparison from the aspects of imaging quality, imaging efficiency, and robustness to noise not only reveals the connections and differences among these illumination schemes, but also shows the advantages, limitation, and suitable applications of their own.