Fundamental Research on Detection and Classification Limits in Spectral Imagery

The aim of this project is to apply empirical analyses and a model-based prediction capability to explore spectral imaging system parameter performance sensitivities and trends with a goal of developing insights into their fundamental limits. Previous years have focused on the model-based prediction capability to identify system limitations. During this fifth and final year of the program, PhD student Muskan Kingrani conducted an experiment as part of the ROCX 2025 field campaign. Lattice based targets developed by previous PhD student Chase Canas were deployed in horizontal to-the-ground and tilted normal-to-the-sun configurations as shown in Figure 1. The targets were imaged by several hyperspectral imagers including RIT's MX-1 platform with a Headwall Nano VNIR camera and the HySpex Mjolnir VNIR/SWIR camera. Figure 2 shows a comparison of receiver operating characteristic (ROC) curves derived from the Headwall Nano imagery using a matched filter with full pixel panels as the target spectrum. The figure demonstrates the higher detection performance achieved for the titled panel due to the higher signal present with the panel normal-to-the-sun.