Study On The Methods Of Target Detection In Polarimetric Sar Imagey

As a type of active microwave imaging sensor that can work at any time day or night and under all weather conditions, Synthetic Aperture Radar (SAR) has become one of indispensable tools to get the important information for all the countries of the world today. Polarimetric Synthetic Aperture Radar (POLSAR), which has the capability of polarimetric measuring, is more favored by all the countries for its more powerful capability in the information getting. With the constant development of POLSAR theory and the increasing of POLSAR data, it has become possible as well as urgent to develop the work of POLSAR imagery interpretation. The dissertation works at the target detection based on the POLSAR imagery that is one part of the imagery interpretation, and the main content contain the following parts:First of all, the basic theory of POLSAR was expatiated, including the description of the polarization of electromagnetic waves, target's depolarization effect and its description, the polarization synthesis and polarization signature.After that, several standard polarization detectors were referred and their performances were evaluated and compared by the Monte Carlo method with the simulated data. Aiming at solving the problem in the parameter estimation of the PWF, two improved PWF methods were then proposed: (1) The improved PWF based on sampled data choice when the uniformity factor was taken in and a quadrant sub-window was designed; (2) The improved PWF based on sampled data choice with the theory of Freeman target decomposition.Finally, Oriented to solving the statistical model mismatching problem while applying traditional 2P-CFAR detection algorithm to polarimetric fusion image, a novel parameter modeling method base on sampled data Segment-Sampled Way combining the Local Modeling Way was proposed and its effectiveness was showed by theory analyse and the experiment. A fine detection result was gotten while applying the improved CFAR method to detect the targets with the model above.