Spectral Correlation Mapper (SCM): An Improvement on the Spectral Angle Mapper (SAM)

The classification methods for feature spectra are based on a comparison of the spectral image with a reference spectrum (endmembers or spectral libraries). The comparison is accomplished through a criterion of similarity. The Spectral Angle Mapper SAM is one of the leading classification methods because it evaluates the spectra similarity in order to repress the influence of the shading to accentuate the target reflectance characteristics (Kruse et al., 1992; Kruse et al., 1993). A lot of studies use SAM for the spectral analysis (Yuhas et al., 1992 e; Conel et al., 1992; Richardson & Kruse, 1999; McCubbin et al., 1998; Riaza, et al., 1998; Bogliolo et al., 1998). In this paper, the SAM formulation is discussed and a new method is proposed for its improvement.