Polarization-based spatial and frequency domain techniques for improved target detection.

摘要

This thesis investigates several spatial and frequency techniques for target detection. These techniques are applied to unpolarized and polarized images. In the spatial domain, we consider Fisher's ratio (FR) and probability density function-based Fisher's ratio (PDF-FR). A comparison is drawn based on the outputs of these two methods. PDF-FR test is shown to have better target detection ability when compared to the FR. In the frequency domain, we consider fringe-adjusted joint transform correlation (FJTC) and wavelet filter-based joint transform correlation (WJTC). A comparison is drawn based on the output of these two methods. The WJTC technique involves more computational complexity but may provide better performance for some scenarios. Performance metrics which are obtained due to change in background color and textures using the FJTC and WJTC techniques are compared.