Figure 2

Principal components analysis. Percent explained variance depicts a clear elbow at PC4, also corresponding to covariance matrix eigenvalue of 1 (purple dashed line), below which, the principal components possess less distinctive features than the original dataset (a). The cumulative explained variance shows that first 4 PCs account for \(74.5\%\) of the variance in the data (b). Biplots of PC2 vs PC1 (c) and PC4 vs PC3 (d), and their insets (e,f, respectively), show the most significant contributing polarimetric and texture parameters in each PC. Acronyms: PD pixel density, SHG SHG intensity, R R-ratio, DCP degree of circular polarization, CD SHG-CD, LD SHG-LD, CON contrast, COR correlation, ENT entropy, ASM angular second moment, IDM inverse difference moment, MN mean, MAD mean absolute deviation.