An Efficient Technique for Hyperspectral Endmember Extraction based on SE

Hyperspectral Endmember extraction of a set of accurateendmembers is critical for the proper unmixing of Hyperspectral image. Several preprocessing algorithms such as spatial preprocessing (SPP), region based spatial preprocessing (RBSPP), and spatial spectral preprocessing (SSPP) have been developed for the extraction of endmembers. These algorithms require complex operations and huge computational effort. Thus, this paper proposes an efficient preprocessing technique for hyperspectralendmember extraction called SE 2 PP which is based on the integration of spatial and spectral information. The proposed approach can be combined with the existing algorithms for endmember extraction reducing the computational complexity of those algorithms while providing similar figures of accuracy.