Advanced hyperspectral imager design for single-device pansharpening.

Abstract

Hyperspectral imaging (HSI) provides unparalleled spectral detail for remote sensing, yet accessible pushbroom imagers often suffer from limited or anisotropic spatial resolution. Super-resolution techniques such as pansharpening can address this; by fusing high-resolution panchromatic images with hyperspectral datacubes, however its typical reliance on external, third-party imagery limits flexibility for small-scale research or unique targeting. To overcome these constraints, this thesis presents the design, development, and application of a compact, dual-modality imaging device, integrating both hyperspectral and high-resolution panchromatic sensors. By simultaneously capturing both modalities, the system enables intra-device pansharpening without external datasets. The co-aligned panchromatic sensor also provides a live video feed for region targeting. The research first explores the optical design of two spectrometers: a custom-lens theoretical model developed to maximise photon throughput and spatial resolution, and a redesign of the open-source OpenHSI architecture, similarly using commercial-off-the-shelf (COTS) components to improve spectral uniformity. The COTS model is physically assembled into a dual-detector setup with custom 3D-printed housing, followed by calibration for wavelength, radiance, and aberration correction. To validate the system, datacubes of contrast targets and natural scenes are captured and processed using a modified Gram-Schmidt Adaptive (GSA) pansharpening algorithm. Comparisons of pre- and post-fusion datacubes confirm significant spatial resolution enhancements while preserving spectral integrity. This work demonstrates the viability of intra-device pansharpening to deliver high spectral resolution alongside isotropic spatial contrast. The findings establish a scalable, cost-effective pathway for advancing compact, dual-modality hyperspectral imaging technologies.