Hyperspectral imaging: history and prospects

Abstract

Hyperspectral (HS) imaging, which acquires the detailed spectral information of an object, has attracted extensive interest in various fields, such as remote sensing, agriculture, and biomedicine. In the dawn of HS imaging, HS imaging relied primarily on time-consuming scanning techniques to achieve high spatial and spectral resolution, limiting its applicability. Subsequently, snapshot (non-scanning) HS imaging emerged, enabling high-speed capturing. The snapshot method captures HS images in a single exposure, but it has the drawback of lower spatial resolution. With recent advancements in computational technology, high-efficiency HS imaging has been pursued in cooperation with image post-processing. In this review, the historical evolution of HS imaging is described focusing on spectroscopic techniques. An up-to-date HS imaging technique, computational HS imaging using a spatial-spectral random coded mask, is introduced with a brief explanation of its working principle. The HS imaging technique demonstrates outstanding characteristics in terms of sensitivity, spatial resolution, spectral accuracy, and frame rate. Several applications of HS imaging are introduced showing improved accuracy of image analysis compared to traditional RGB image analysis. Finally, the remaining challenges and prospects are discussed.