Abstract:
|
Raman spectroscopy plays a crucial role ranging from fundamental science, nano-materials, and medical applications. However, background signals can heavily interfere with the analysis of Raman spectra. Thus, we perform baseline correction on raw spectra in order to eliminate the unwanted background signals. Performance of traditional baseline correction methods can be affected by choice of cost function, number of tuning parameters, or parameter values. For example, polynomial fitting may not sufficiently capture complicated baselines due to its small number of tuning parameters. We propose a novel approach, Iterative Smoothing-spline with Root Error Adjustment (ISREA), which employs an asymmetric root error function. ISREA is relatively simple to implement, converges quickly, and accurately captures baseline of the spectrum. We compare our method with another baseline correction method, Goldindec, on both simulated and real data. We show ISREA has better overall performance.
|