Abstract
Linear and nonlinear components analysis of data from a monostatic laser polarimeter is developed and applied to the task of remote, non-imaging discrimination among different textures on paint and polymer coupons independent of their spatial orientations. Both principal-components analysis and nonlinear components analysis are applied to multi-dimensional laser data in measured Mueller matrices, with discrimination via cluster segmentation in derived linear and nonlinear constant channels. Textures on the discriminated coupons are generated by heating and illustrated in optical micrographs.