Building End Member Hybrid Profiles from Hyper Spectral Images for Unsupervised Land Cover Mapping

Abstract

The end member hybrid profile (EMHP) representing end individuals extracted from multispectral photos (MSI) and spectral libraries has been delivered for unsupervised land cowl mapping. Compared to traditional unsupervised land cover mapping techniques, EMHP correctly reduces the records loss compared to digital numbers (DNs) by maintaining the spectral library, and MSI ceases members independently. This advancement can improve mapping accuracy substantially. Furthermore, EMHP can represent more details than traditional mapping gear because of the potential to assemble cease individuals from hyperspectral images (HSI). The cease contributors from the HSI include more spectral facts than MSI and feature the ability to represent land covers in each vicinity accurately. These blessings make EMHP a promising approach for unsupervised land cover mapping. However, computational value and a wide variety of quit members produced from the HSI want to be addressed for this method to be extra powerful in applications.