Polarimetric Sar Image Enhancement And Classification Techniques

At present the polarimetric Synthetic Aperture Radar (SAR) system is an important aspect of the microwave imaging research and application. It provides a 2 by 2 complex scatter matrix of a target, not only amplitude but also phase. It can obtain times of information than traditional radar. Therefore, polarimetric SAR can efficiently enhance the ability to acquire the information of targets, and provide an effective approach to further analyze, identify and detect targets. Because of difficulties in making a polarimetric radar system, there is no progress in the field of radar polarimetry for a long time. As the sophisticated theory coming and electronic technology developing, radar polarimetry has become an indispensable tool, and polarimetric radars have been applied in several fields such as military, remote sensing, agriculture, hydrology and monitor, and so on. Polarimetric SAR image enhancement and classification technologies are investigated in this thesis. The mainly contents are described as followings:1. We analyze the important application of polarmetric synthesis technology in polarmetric SAR image processing: two kinds of target contrast enhancement and two targets null in polarmetric SAR image. The linear-weighted method and numerical method of Sequential Unconstrained Minimization Technique (SUMT) of contrast enhancement can enhance the contrast of two targets in polarmetric SAR image. Two targets null can eliminate two targets which we don't hope appear in polarmetric SAR image.2. The span, optimal weighting and polarimetric Whitening Filter methods have disadvantages of losing polarimetric information. We research two methods of speckle filtering in polarimetric SAR image: speckle filtering based on the eigenvalue-decomposition and adaptive speckle filtering in polarimetric SAR image. The experiment results indicate two methods can preserve the polarimetric information in polarimetric SAR data. The processed data can be used for other polarimetric information processing fields, such as polarmetric synthesis, polarmetric decomposition, target recognition, target classification and so on. Because the data have a little speckle, the obtainable polarimetric information may remarkably increase.3. We research three methods of polarmetric SAR image classification. The first method classify target in nine classes using two parameters H andαwhich are obtain from eigenvalue-decomposition of polarmetric coherency matrix. The classified results accord with actually case and reflect the scatter characteristic of actually targets. The second method is maximum likelihood iterative classifier based on the complex Wishart distribution for the polarimetric covariance matrix. Because the second method need use inverse matrix and logarithm frequently, it has more computing. We introduce the concept of target polarmetric difference degree. The third method is iterative classifier of polarimetric SAR images based on target polarmetric difference degree. The experiment results indicate the second and third methods get an excellent classification result.