Abstract

All quantitative data evaluation techniques applied to spectroscopy are based on the assumption that the baseline is stable in time. If this prerequisite is violated, major concentration errors can result since drifts are evaluated along with true spectroscopic features. For handling baseline drifts two improved principal component regression (PCR) methods are presented and compared to conventional Savitzky–Golay preprocessing followed by conventional PCR. The proposed drift-correction methods take advantage of baseline drifts being rather broad compared to the absorption features. The only assumption made is that drift effects can be modeled sufficiently by polynomials of user-selectable order. One correction method modifies principal components such that drifts of polynomial shape are orthogonal to the calibration model and thus cannot influence the concentration result. The second method extends the calibration model by synthetic so-called pseudo-principal components. While the principal components model the true spectral features, the pseudo-principal components describe drifts simultaneously and independently. Hence, drifts are explicitly included into the calibration and cannot cause erroneous concentration results. It is demonstrated that both correction methods are equivalently as powerful as the conventional PCR in the absence of drifts and superior if drifts are present. The overall performance—in the absence and presence of baseline drifts—of the novel methods makes them more versatile and reliable than the conventional Savitzky–Golay data preprocessing. In almost all investigated cases, the average concentration errors were significantly smaller than those obtained by Savitzky–Golay preprocessing. Furthermore, polyPCR and pPCR do not need laborious optimizations as Savitzky–Golay does for preventing suppression of relevant signal components. polyPCR demands less computation expense than Savitzky–Golay, and pPCR extracts the drift spectrum as additional qualitative information not provided by Savitzky–Golay.

Drift correction in a multichannel integrated optical Young interferometer

Aurel Ymeti, Jan Greve, Paul V. Lambeck, Robert Wijn, Rene G. Heideman, and Johannes S. Kanger
Appl. Opt. 44(17) 3409-3412 (2005)

Precision evaluation of Raman densimetry for carbon dioxide: improvement by correction for the drift effect

Junji Yamamoto and Yuuki Hagiwara
Appl. Opt. 63(5) 1402-1410 (2024)

Sub-nanometer drift correction for super-resolution imaging

Y. Tang, X. Wang, X. Zhang, J. Li, and L. Dai
Opt. Lett. 39(19) 5685-5688 (2014)

Optical resonant sensors: a method to reduce the effect of thermal drift

Thanh Le, Anatoliy Savchenkov, Nan Yu, Lute Maleki, and W. H. Steier
Appl. Opt. 48(3) 458-463 (2009)

Using fixed fiduciary markers for stage drift correction

Sang Hak Lee, Murat Baday, Marco Tjioe, Paul D. Simonson, Ruobing Zhang, En Cai, and Paul R. Selvin
Opt. Express 20(11) 12177-12183 (2012)