The Directly-Georeferenced Hyperspectral Point Cloud: Preserving the Integrity of Hyperspectral Imaging Data

The raster data model has been the standard format for hyperspectral imaging (HSI) over last four decades. Unfortunately, it misrepresents HSI because pixels are not natively square nor uniformly distributed across imaged scenes. To generate end products as rasters with while preserving spectral integrity, nearest neighbor resampling methodology is typically applied. This process compromises spatial integrity from original shifted, duplicated and eliminated so that can conform to structure. Our study presents a novel point cloud representation preserves spatial-spectral of more effectively than conventional pixel rasters. Directly-Georeferenced Hyperspectral Point Cloud (DHPC) generated through fusion workflow be readily implemented into existing processing workflows used by providers. effectiveness DHPC shown datasets. These datasets were collected at three different sites two sensors captured information each site various resolutions (ranging ∼1.5 cm 2.6 m). was assessed based on quality metrics (i.e., loss, duplication shifting), storage requirements applications. All studied characterized either substantial loss (∼50–75%) or (∼35–75%), depending resolution grid in methodology. Pixel shifting ranged 0.33 1.95 pixels. zero shifting. Despite containing additional surface elevation data, up 13 times smaller file size corresponding Furthermore, consistently outperformed all tested applications which included classification, spectra geo-location target detection. Based findings this work, developed potential push limits distribution, analysis application.