Novel Computational Phase Imaging Techniques

Phase imaging, or more precisely, imaging relative differences in optical path length, is an important tool with many applications. For example, many biological specimens of interest are nearly transparent, and phase imaging is one method by which such specimens can be imaged noninvasively with high contrast. Recent developments in optimization as well as ever-increasing computational power enable novel phase imaging techniques based on solving inverse problems modeled directly after first principles. For example, the factored form descent solver from coherence retrieval can be extended to explicitly enforce coherence and thus can be used for phase retrieval from measurements through arbitrary (known) optical systems.


(a)

(b)


(c)

(d)

(e)
An inverse problem approach using shifted wavefront sensor images (i.e. light field images of a coherent source) yields the phase profile of a hypergeometric Gaussian (HyGG) beam (a) and a human cheek cell (b). The former is from simulated noisy measurements, and the latter is from experimentally captured images. Both images yielded details finer than the lens array pitch, which was roughly 1/16 of the width of the image in (a) and 1/32 of the width of the image in (b). A different phase imaging technique combining wavefront sensing and the transport of intensity equation (TIE) method yields a high quality phase profile reconstruction of beads over an aberrated wavefront (c), combining high frequency information from the TIE method (d) and low frequency information from the wavefront sensor (e).

Relevant Articles and Conference Proceedings