Determining the index of refraction of an unknown object using passive polarimetric imagery degraded by atmospheric turbulence

In this research, an algorithm is developed to estimate the index of refraction of an unknown object using passive polarimetric images degraded by atmospheric turbulence. The algorithm uses a variant of the maximum-likelihood blind-deconvolution algorithm developed by LeMaster and Cain to recover the true object (i.e., the first Stokes parameter), the degree of linear polarization, and the polarimetric-image point spread functions. Nonlinear least squares is then used to find the value of the complex index of refraction which best fits the theoretical degree of linear polarization, derived using a polarimetric bidirectional reflectance distribution function, to the turbulence-corrected degree of linear polarization. To verify the proposed material-characterization algorithm, experimental results of two painted metal samples are provided and analyzed. Possible uses of this novel algorithm include intelligence-gathering and nondestructive inspection/evaluation applications such as corrosion and crack detection/characterization.;Before the algorithm described above is implemented and experimentally verified, the results of two intermediate research steps are provided and discussed. The purpose of the first research step is to verify the accuracy of the polarimetric bidirectional reflectance distribution function chosen for this research. This is accomplished by comparing predictions made using the model to exact electromagnetic solutions of a rough perfect-reflecting surface and to experimental Mueller matrices of two rough metallic samples. In the second research step, the polarimetric bidirectional reflectance distribution function is used to create two material-classification algorithms which use turbulence-degraded polarimetric imagery. The first algorithm is a dielectric/metal classifier. It uses the blind-deconvolution algorithm mentioned above to remove atmospheric distortion and correctly classify the unknown object. The second classification algorithm, an enhanced version of the first, determines whether an object is composed of aluminum, iron, or dielectric materials. This enhanced material classification provides functional information about the unknown object. Experimental results of two dielectric and metallic samples are provided to validate the proposed classification algorithms. The results of these analyses are presented and discussed in this dissertation.;After results of the two intermediate research steps and the material-characterization algorithm have been analyzed, this dissertation is concluded. A summary of the work performed in this dissertation and a discussion of possible future research areas related to this work are provided.