This research paper introduces a novel synthetic hyperspectral dataset that overcomes the limitations of relying on a single camera for high spectral and spatial resolution imaging. The dataset combines three modalities: RGB, push-broom visible hyperspectral camera, and snapshot infrared hyperspectral camera, each offering unique spatial and spectral resolutions. While RGB cameras provide high spatial resolution but limited spectral resolution, hyperspectral cameras offer high spectral resolution but at the expense of spatial resolution. Additionally, different hyperspectral cameras have varying capturing techniques and spectral ranges, making data acquisition complex. By integrating the photometric properties of these modalities, a single synthetic hyperspectral image is generated, enabling the exploration of spectral-spatial relationships for enhanced analysis, monitoring, and decision-making in various fields. The paper emphasizes the significance of multi-modal fusion in producing a high-quality synthetic hyperspectral dataset with consistent spectral intervals between bands.
A Synthetic HyperSpectral Dataset with High Spectral Spatial Resolution Achieved through Multi-Source Fusion. (arXiv:2309.00005v1 [cs.CV])
