Presentation: 2025 ND EPSCoR Annual conference 

October 21, 2025, NDSU Memorial Union, Fargo, North Dakota

Real-time Data Processing of Snapshot Hyperspectral Imaging System for Biomedical Applications

We introduce a real-time data processing framework for microlens-array-based snapshot hyperspectral cameras, which capture hyperspectral datacubes in a single camera exposure. The framework consists of a one-time pre-indexing phase and a real-time reconstruction engine. In the pre-indexing stage, we perform both spectral and spatial calibration across the detector to generate a lookup table. This includes robust geometric fitting of the hyperpixel grid and precise spectral trace alignment. A sub-pixel-accurate grid extraction algorithm is used to regularize the spatial-spectral layout across the focal plane. During real-time operation, each acquired frame is rapidly transformed into a hyperspectral datacube through an interpolation-based reconstruction process. The system remaps the 2D spectral data into a 3D hyperspectral cube using the precomputed grid geometry, applying interpolation in both spatial and spectral dimensions. Our computational pipeline supports video-rate processing, making the system suitable for a wide range of biomedical applications.