ghost-imaging

Self-supervised deep convolutional reconstruction for ghost imaging.

This repository introduces Generalized SHSI using tunable sparse Hadamard speckle patterns to achieve noise‑robust, high‑resolution imaging with low‑order Hadamard matrices while alleviating the curse of dimensionality.

This repository presents loop differential ghost imaging with line patterns generated by a 1D OPA with a grating waveguide, revealing a limitation of conventional ghost imaging caused by noise‑symmetry breaking under such illumination.

Few codes with the tools needed for doing Ghost Imaging simulations: speckle generation, simulation of measurements, recovery, etc.

Ten deep-learning single-pixel image reconstruction methods benchmarked on one shared Hadamard acquisition (PyTorch)

Image-free segmentation straight from single-pixel measurements — content-adaptive lift + task-prioritized loss scheduling (PyTorch)

FPGA-based computational ghost imaging system on the Nexys Video Artix-7 — reconstructs images from a single-pixel detector using structured light patterns, with on-chip acquisition and reconstruction. Part of an instrumentation platform alongside Ramsey.