Discrimination of Objects Within Polarimetric Using Principal Component and Cluster Analysis

摘要

We apply two multivariate image analysis techniques to several sets of spatially coincident, long-wave infrared (LWIR), polarimetric images to improve target contrast and/or aid in the identification of certain object features not present in the original image set. Principal component analysis (PCA) was used to obtain new representations of the scene that highlight target features based upon the variance of the variables used in the analysis. We show a representation of the scene that maintains the same level of target-to- background contrast (when compared to the conventional thermal and degree of linear polarization images), as well as additional information content contained in the resultant PCA analysis. Cluster analysis (CA) was used to group pixels of the image that have similar values for the variables chosen. We show that this method is an effective means for separating objects of interest from complex backgrounds, as well as subdividing different features of an object.