A Simple Approach to Analysis Temperature-dependent Terahertz Spectra Based on Zernike Moments

The spectral fingerprint in terahertz frequency region shows a remarkable shift with temperature. That means if a qualitative and quantitative model has been trained for samples at constant temperature, it may not always work for an arbitrary temperature. To solve this problem we introduce Zernike image moment theory and show how it may be used for classifying and quantifying the terahertz spectra, isolating the influence of temperature variation.