GEOMScope: Large Field-of-View 3D Lensless Microscopy with Low Computational Complexity

This is a temporary repository for work
Tian, F., Hu, J., Yang, W., GEOMScope: Large Field-of-View 3D Lensless Microscopy with Low Computational Complexity. Laser & Photonics Reviews 2021, 15, 2100072.

Paper

Clone this repository:

git clone https://github.com/Fengshub/GEOMScope

Ray tracing algorithm and clustering algorithm for sparse objects

We provided the code for 3D object reconstruction from measured image with ray tracing. The ray tracing algorithm is to generate for intermediate 3D reconstruction results before post processing with CNN or clustering algorithms. The raw measurement data example is provided in folder, with camera TIF image in drive folder.
The algorithms includes data loading, PSF initialization, ray tracing and multi-threshold clustering.
Our random microlens array has each lens unit coordinates recorded in mm units, the coordinates are used to initialize point spread function (PSF).

CAD files and macros

We provide CAD files of designed microlens array mold for 3D printing. We also provide lens array & emission filter holder we used, and several CAD molds for imaging targets.

UNet model for post processing

We provide code to build convolutional network modules for post processing, and training.

Zemax Ray tracing simulation

We provide ZemaxOpticStudio lens zmx file in non sequential mode on the full microlens array simulation with our CAD target.