CFAR ship detection system using polarimetric data

The objective of this work is to develop optimal polarimetric Constant False Alarm Rate (CFAR) ship detection system. Polarimetric transformation/decompositions, clutter analysis, modeling, Principal Component Analysis (PCA), and multi-CFAR detection are the necessary components of optimal polarimetric CFAR ship detectors. The resulting CFAR detector outperforms the conventional polarimetric CFAR detector by providing higher probability of detection without introducing much more computations. Optimal polarimetric CFAR detection procedures are proposed in this report. Given the simulated polarimetric RADARSAT-2 data, different polarimetric transformations and decompositions are applied. The resulting images will be transmitted to an adaptive Principal Component Analysis (PCA) block. Through the adaptive PCA block, the image of the first principal component which has the highest SNR among all the images (including the original, transformed/decomposed images, and the images after the adaptive PCA) will be used for ship detection. Optimal multi-CFAR detection will be applied to this image and then the final decision will be made.