Abstract
The majority of hyperspectral data exploitation algorithms are developed using statistical models for the data that include sensor noise. Hyperspectral data collected using charge-coupled devices or other photon detectors have sensor noise that is directly dependent on the amplitude of the signal collected. However, this signal dependence is often ignored. Additionally, the statistics of the noise can vary spatially and spectrally as a result of camera characteristics and the calibration process applied to the data. Here, we examine the expected noise characteristics of both raw and calibrated visible/near-infrared hyperspectral data and provide a method for estimating the noise statistics using calibration data or directly from the imagery if calibration data is unavailable.
© 2011 Optical Society of America
Full Article | PDF ArticleMore Like This
Torbjørn Skauli
Opt. Express 19(14) 13031-13046 (2011)
Jaka Katrašnik, Franjo Pernuš, and Boštjan Likar
Appl. Opt. 52(15) 3526-3537 (2013)
Peng Fu, Changyang Li, Yong Xia, Zexuan Ji, Quansen Sun, Weidong Cai, and David Dagan Feng
Appl. Opt. 53(30) 7059-7071 (2014)