This project aims to build a new vision model that reconstructs objects from their in-line holograms. The objective is to minimize the twin-image effects and image artifacts (aberrations, speckles, sensor effects, etc.).

A robust, translation invariant architecture for generative image watermarking with provable resistance to cross-image codebook subtraction

Fourier optics - Hologram fabrication