Hyperspectral and multispectral data fusion using fast discrete curvelet transform for urban surface material characterization

The objective of the present study is to analyze the quality of hyperspectral data fusion using low spatial hyperspectral (LSH) Airborne Visible InfraRed Imaging Spectrometer-Next Generation (AVIRIS-NG) 8 m data and high spatial multispectral (HSM) WorldView-3 image at 1.24 m remote sensing images with spectral unmixing technique. The resultant HSH data shows new prospects for urban surface material characterization with spectrally distinct classes. The spatial resolution of LSH is enhanced by injecting the high-frequency details from the corresponding HSM bands in fast discrete curvelet transform domain. The image fusion-based products' quality has been analyzed by endmembers extraction and fractional maps generated using Piecewise Convex Multiple-Model Endmember Detection (PCOMMEND) method. Experimental results showed that the fusion has improved the spatial as well as spectral separability to extract the endmembers, particularly for the urban surface materials like the combination of water and asphalt, and bare soil and roof tiles.