Mixed and Sub-Pixel Target Detection Using Space Borne Hyper-Spectral Imaging Data: Analysis and Challenges

In the present study, exploration of HySIS data is carried out for performance analysis in terms of image data quality and spatial/spectral characteristics in order to identify and characterize the targets. The study highlights the different aspects related to HySIS sensor and pre-processing. The performance of HySIS data vis-à-vis PRISMA data is also carried out as both the sensors has 30m spatial resolution. It is found that PRISMA data could represent the field target reflectance but the HySIS data entirely differ with the field target spectra particularly in SWIR region. Identification and characterization of target material is explored by conducting field experiments in synchronization with HySIS over-passes for different scenarios such as sub-pixel and mixed pixel. Spectral measure method and adaptive coherence estimator (ACE) method are used for sub-pixel and mixed pixel target detection respectively using HYSIS data. Sub-pixel target was detected with second highest spectral measure score of 0.36. For mixed pixel case, the ACE method score was highest (0.5) for the target of interest. The study highlights the target detection capabilities of HYSIS sensor along with image quality characteristics necessary for data processing.