Reflective coherent diffraction imaging with binary random sampling and updated support constraints

Coherent diffraction imaging (CDI) has emerged as a thriving field promising applications in materials and biological sciences. Transmission geometry is adopted in most CDI methods, which is not suitable for opaque structures or objects of interest comprising only surfaces or interfaces. Here, we present a reflective CDI system using binary random sampling patterns, which extends the application range to bulk and opaque samples. Instead of resorting to an array of lenslets in Shack–Hartmann wavefront sensing or a reference beam in interferometry, the spatial information is captured by illuminating the object with four patterns generated with a digital micromirror device (DMD). On this basis, an adaptive algorithm for updating the support constraints in the iteration process is proposed to solve the problem of pattern deformation caused by the height variations of the sample surface. Our method achieves high-speed modulation, high-accuracy and high-fidelity quantitative phase imaging (QPI). The effectiveness of the proposed method is demonstrated by both simulated and real experiments.