Giters

azavea / nasa-hyperspectral

An event-driven image processing pipeline for developing our foundational capability to work with HSI data sources.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Implement robust endmember selection

jpolchlo opened this issue · comments

commented

As an entrypoint into endmember selection, we should investigate the robust successive projection algorithm (RSPA). This is a means to determine, in a noise-tolerant way, a set of spectral endmembers that describe the dominant spectral components present in a sample of spectra. These can then be used in unmixing operations to disaggregate the pixels of an image according to the abundance of each endmember's material.

References:
Gillis, N. (2019, December). Successive projection algorithm robust to outliers. In 2019 IEEE 8th International Workshop on Computational Advances in Multi-Sensor Adaptive Processing (CAMSAP) (pp. 331-335). IEEE.