Hyperspectral vision beyond 3D: a review

Abstract

Hyperspectral imaging (HSI) technique has evolved dramatically over the past few decades, demonstrating its capability for a wide range of applications. Combining HSI with other advanced imaging modalities have resulted in versatile systems with exceptional imaging and spectroscopic capabilities. HSI originated as an imaging technique for assessing the spatial and spectral attributes of the objects under inspection. Attributes extracted from HSI can be well represented and visualised in 3-dimensional (3D) space using virtual reality (VR) and augmented reality (AR). Later advancements led to the development of 4D HSI cameras that can track the spatial and spectral characteristics of objects over time and became a hot topic in precision agriculture, medical imaging, and optical sorting. Recently, 5D HSI systems utilizing structured light were developed for measuring the time-dependent spectral characteristics along the entire shape of macroscopic objects under investigation, providing excellent spectral, spatial, and axial resolutions. Latest compact HSI cameras with panoramic imaging capabilities offering high spatial and spectral resolutions are perfect candidates for remote sensing applications. Such capabilities facilitated the evolution of HSI into machine vision, resulting in the creation of a new realm named hyperspectral vision that enables intelligent automation. This review highlights the technological advancements in the field of HSI, focusing on advanced imaging modalities to realise spectral vision beyond 3D for applications in machine vision and Industry 4.0. The review also throws light on the challenges faced during real-life implementations of such technologies and possible mitigation strategies. Approaches for enhancing the HSI data using matured technologies such as AR and VR are also discussed.