“…Some popular preprocessing techniques, including a derivative method (Karstang and Kvalheim, 1991), Fourier filtering (Arnold and Small, 1990), and orthogonal signal correction (Svensson et al, 2002), have been used to remove background signals and noise from spectral data. Recently, wavelet transform methods have been widely used as alternative preprocessing tools in VIS/NIR spectral analysis (Walczak and Massart, 1997;Jetter et al, 2000). Studies have demonstrated that a discrete wavelet transform (DWT) is an excellent tool for fine-and coarse-scale spectral feature separation in hyperspectral data (Bruce and Koger, 2002;Zhang et al, 2005).…”