Ptychographic Imaging with a Fiber Endoscope via Wavelength Scanning

Ptychography is a phase retrieval technique that has recently seen a resurgence in optical imaging applications thanks to an increase in computing power and memory [1]. These advancements have also facilitated the use of deep neural networks (DNNs) for addressing inverse problems in computational imaging [2]. In the present paper, we combine these recently established techniques to create a hybrid model of ptychography and neural networks which introduces an alternative endoscopic method [3]. Two fibers are used, one for illumination and one for collection of the diffracted wavefront. Wavelength scanning through single mode fiber has been used instead of the more commonly used mechanical scanning ptychography method to achieve a simpler design and reduce positioning errors. Neural networks are used to compensate for the wave distortion due to propagating through the multicore collector fiber.