Abstract

Cosmic rays can degrade Raman hyperspectral images by introducing high-intensity noise to spectra, obfuscating the results of downstream analyses. We describe a novel method to detect cosmic rays in deep ultraviolet Raman hyperspectral data sets adapted from existing cosmic ray removal methods applied to astronomical images. This method identifies cosmic rays as outliers in the distribution of intensity values in each wavelength channel. In some cases, this algorithm fails to identify cosmic rays in data sets with high inter-spectral variance, uncorrected baseline drift, or few spectra. However, this method effectively identifies cosmic rays in spatially uncorrelated hyperspectral data sets more effectively than other cosmic ray rejection methods and can potentially be employed in commercial and robotic Raman systems to identify cosmic rays semi-autonomously.

© 2019 The Author(s)

Single-shot chemical detection and identification with compressed hyperspectral Raman imaging

Jonathan V. Thompson, Joel N. Bixler, Brett H. Hokr, Gary D. Noojin, Marlan O. Scully, and Vladislav V. Yakovlev
Opt. Lett. 42(11) 2169-2172 (2017)

Detection and correction of spectral and spatial misregistrations for hyperspectral data using phase correlation method

Naoto Yokoya, Norihide Miyamura, and Akira Iwasaki
Appl. Opt. 49(24) 4568-4575 (2010)

Tensor methods for hyperspectral data analysis: a space object material identification study

Qiang Zhang, Han Wang, Robert J. Plemmons, and V. Pau'l Pauca
J. Opt. Soc. Am. A 25(12) 3001-3012 (2008)

Parametric level set reconstruction methods for hyperspectral diffuse optical tomography

Fridrik Larusson, Sergio Fantini, and Eric L. Miller
Biomed. Opt. Express 3(5) 1006-1024 (2012)

Semi-synthetic data generation to fine-tune a convolutional neural network for retrieving Raman signals from CARS spectra

Ali Saghi, Rajendhar Junjuri, Lasse Lensu, and Erik M. Vartiainen
Opt. Continuum 1(11) 2360-2373 (2022)