Fusion of IIRS and M3 data for the purpouse of finer resolution mineral mapping

Hyperspectral images, despite their rich spectral information, often suffer from low spatial resolution due to physical constraints in imaging sensors. However, when higher spatial resolution data of the same scene are available, we can perform data fusion to generate hyperspectral images with high spatial resolution. This fused data can be viewed as the output of a synthetic sensor that combines the high spatial and spectral resolution data acquired by different sensors. This fusion allows for new applications with increased accuracy, such as high-resolution mapping of minerals and surface materials. Imaging spectroscopy facilitates the identification and discrimination of materials and their constituents. Data fusion enhances both the spatial and spectral characteristics of the initial data. It is based on the synergistic exploitation of data from different sources, aiming to produce superior results. By integrating data from The Moon Mineralogy Mapper (M3) by NASA and the Imaging Infrared Spectrometer (IIRS) by ISRO, we can improve the spatial and spectral resolutions, enhance measurement accuracy, and reduce uncertainties. This will enable a more precise assessment of the mineral composition of the area of interest. The objective is to fuse high spatial resolution data, which has discontinuities in the spectral domain, with low spatial resolution data that has continuous spectra. The ultimate goal is to estimate an image with high spatial and spectral content, providing a more comprehensive and accurate understanding of the area of interest. We replaced the noisy bands in the M3 and IIRS data and used cubic convolution to resample the M3 bands to the IIRS band’s native spatial resolution. However, the M3 bandwidth is different from the IIRS bandwidths. Nevertheless, this gap-filling procedure will allow us to identify endmembers. As a followup study we are going to employ a spectral unmixing technique to obtain endmembers information and high-resolution abundance matrices from the initial images. Data fusion helps overcome the limitations of individual datasets, exploit the strengths of different sensors, and extract more valuable information.