Alex Robey '18, Engineering and Mathematics

Computationally Expediting Fourier Pytchographic Microscopy

Fourier Ptychographic Microscopy (FPM) is a computational imaging technique that overcomes the limitations of an optical system to generate a super-resolution image by iteratively stitching together low-resolution, variably-lit images in Fourier space. However, FPM requires hundreds or thousands of low-resolution images; the process of collecting and processing these images often makes this technique infeasible. In this work, we introduce a deep-learning architecture that computationally optimizes the illumination patterns used to generate the low-resolution images needed for super-resolution image generation. We demonstrate that our algorithm successfully determines illumination patterns that much more efficiently reconstruct high-resolution images comparable to those created by stitching together thousands of images.