TY - GEN
T1 - Estimating hyperspectral backgrounds
T2 - 38th Annual IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2018
AU - Faybish, O.
AU - Rotman, S. R.
N1 - Publisher Copyright:
© 2018 IEEE
PY - 2018/10/31
Y1 - 2018/10/31
N2 - In this paper, we compare two of the methods to evaluate backgrounds for hyperspectral point target detection, i.e., the mean and the median of the surrounding pixels, and the effect of two types of noise, i.e., spatial and spectral. We found that while the median often technically represents a more accurate estimation based on the mean squared error, the mean produces better results in practice; it maintains spectral coherence in the environment of spatial noise, while the median loses such coherence, especially on edges.
AB - In this paper, we compare two of the methods to evaluate backgrounds for hyperspectral point target detection, i.e., the mean and the median of the surrounding pixels, and the effect of two types of noise, i.e., spatial and spectral. We found that while the median often technically represents a more accurate estimation based on the mean squared error, the mean produces better results in practice; it maintains spectral coherence in the environment of spatial noise, while the median loses such coherence, especially on edges.
UR - http://www.scopus.com/inward/record.url?scp=85063146605&partnerID=8YFLogxK
U2 - 10.1109/IGARSS.2018.8517862
DO - 10.1109/IGARSS.2018.8517862
M3 - Conference contribution
AN - SCOPUS:85063146605
T3 - International Geoscience and Remote Sensing Symposium (IGARSS)
SP - 2785
EP - 2788
BT - 2018 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2018 - Proceedings
PB - Institute of Electrical and Electronics Engineers
Y2 - 22 July 2018 through 27 July 2018
ER -