Linear and nonlinear components analysis of data from a monostatic laser polarimeter is developed and applied to the task of remote, nonimaging 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 multidimensional 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.
© 2007 Optical Society of America
Equations on this page are rendered with MathJax. Learn more.