High Spectral Spatial Resolution Synthetic HyperSpectral Dataset form multi-source fusion
This provides a dataset for improved analysis in fields like remote sensing, but it is incremental as it builds on existing multi-modal fusion techniques.
The paper tackles the challenge of obtaining high spectral and spatial resolution in hyperspectral imaging by creating a synthetic dataset that fuses data from RGB, push-broom visible hyperspectral, and snapshot infrared hyperspectral cameras, resulting in a comprehensive representation with consistent spectral intervals.
This research paper introduces a synthetic hyperspectral dataset that combines high spectral and spatial resolution imaging to achieve a comprehensive, accurate, and detailed representation of observed scenes or objects. Obtaining such desirable qualities is challenging when relying on a single camera. The proposed dataset addresses this limitation by leveraging three modalities: RGB, push-broom visible hyperspectral camera, and snapshot infrared hyperspectral camera, each offering distinct spatial and spectral resolutions. Different camera systems exhibit varying photometric properties, resulting in a trade-off between spatial and spectral resolution. RGB cameras typically offer high spatial resolution but limited spectral resolution, while hyperspectral cameras possess high spectral resolution at the expense of spatial resolution. Moreover, hyperspectral cameras themselves employ different capturing techniques and spectral ranges, further complicating the acquisition of comprehensive data. By integrating the photometric properties of these modalities, a single synthetic hyperspectral image can be generated, facilitating the exploration of broader spectral-spatial relationships for improved analysis, monitoring, and decision-making across various fields. This paper emphasizes the importance of multi-modal fusion in producing a high-quality synthetic hyperspectral dataset with consistent spectral intervals between bands.