PIFPM data

Data:

We provide a high dynamic range (HDR) image dataset for Fourier Ptychographic Microscopy of a Hirudo transverse section (leech), comprising 293 images with a resolution of 5634 × 3753 pixels. The images are named according to their corresponding LED indices in the x and y directions. Note that (x, y) = (15, 15) corresponds to the central LED.

The dataset is accompanied by a dataloader.py script and a setup_parameters.json file. The dataloader.py script is intended to facilitate loading the HDR images, while the setup_parameters.json file stores the relevant system parameters required for reconstruction.

Preprocessing description:

Raw Bayer images are demosaiced using the Adaptive Homogeneity Directed algorithm and then merged into an HDR image. 

Setup parameters:

The HDR dataset was captured and processed by John Meshreki [1] at the CSE Optics Lab, University of Siegen. The illumination unit consists of a programmable Adafruit 32×32 LED array positioned at a distance of 66.5 mm from the object plane, with LEDs spaced 4 mm apart and controlled via an Arduino microcontroller to emit green illumination at a wavelength of 519 nm. The optical system employs a 10x, 0.3 NA infinity-corrected Nikon objective lens (N10X-PF) paired with a 200 mm focal-length tube lens (TTL200-A). These experimental and system-specific parameters are provided in setup_parameters.json for convenient reference and direct use in reconstruction pipelines.

[1] John Meshreki, Jan Philipp Schneider, Onofre Martorell, Michael Moeller, and Ivo Ihrke, "PI-FPM: pupil initialization for Fourier ptychographic microscopy directly from measurement data," Opt. Express 34, 16575-16589 (2026)