SIFT‐Based Automated Registration of Chandrayaan‐2 IIRS Hyperspectral Images

The Imaging Infra-Red Spectrometer (IIRS) is the most advanced reflectance spectrometer currently orbiting the Moon. IIRS was launched on-board Chandrayaan-2 in 2019 to image the lunar surface in the wavelength range of 0.8 to 5.0 µm in 250 contiguous bands at a high spatial resolution of ~80 m/pixel and a spectral resolution of 20-25 nm. The IIRS strips are available in the PDS4-compliant QUB file format. However, the data lack inherent map-projection information. This study presents and implements a different approach to automatically seleno-reference the images obtained from IIRS. Using the SIFT (Scale-Invariant Feature Transform) algorithm, matching common points from the individual resampled pixels of IIRS and LRO-WAC (Lunar Reconnaissance Orbiter - Wide Angle Camera, which has been used as a reference image) are obtained. Our results show that SIFT is able to both identify and match corresponding pixels from both IIRS and WAC with RMS errors < the size of a single IIRS pixel. Hence, any user interested to work with IIRS data may refer to this technique to simplify the registration process of the IIRS strips to their actual ground coordinates.