Research On Polarimetric SAR Image Classification

Polarimetric Synthetic Aperture Radar(POLSAR) is a multi-channel and multi-parameter imaging radar system.By transmitting and receiving different polarized signals of the target,it’s useful for classification of polarimetric SAR images to measure the full polarization scattering echo of the target to gain the features.With the development of various classification for polarimetric SAR images,such as Wishart classifier,H-α-Wishart unsupervised classifier,Freeman-Durden decomposition,polarimetric radars have been applied in many fields such as military,monitor,agriculture,remote sensing,and so on.In this thesis,the description method for polarization state of electromagnetic wave is presented.It’s necessary to set forth concepts and applications of Jones vector and Stokes vector.Then scattering Matrix,scattering vector,covariance Matrix and coherence Matrix are derived.The sorts and meanings of basic scattering mechanisms are studied.Then the pixels are divided into three scattering categories: surface scattering,body scattering and double scattering.It is the groundwork for writing.H-α decomposition is one of the most famous methods for classifying polarimetric SAR images.By using this method,polarimetric SAR image can be divided into two parameters: scattering entropy and average scattering angle.We can classify target into nine classes.The classified results reflect the scatter characteristic of actually targets and accord with actually case.However,target classes are confused in the process of H-α decomposition.So,it’s necessary to make anisotropy to join in the process of classification,then the difference between H-α decomposition and H-α-Wishart a decomposition become visible.In view of classification results filled with speckle,we research two methods of speckle filtering in polarimetric SAR image.The experiment results indicate two methods can preserve the polarimetric information in polarimetric SAR data.It will be investigated the Freeman decomposition to discover the advantages and disadvantages of the decomposition in the article.According to the three scattering powers,the first classification will be achieved.And then,to improve the Freeman decomposition accuracy,it’s necessary to update the classification results based on a complex Wishart classifier.The method has advantages of low time complexity and high convergence speed.At last,a new decomposition method will be proposed to solve the problem of Freeman decomposition which combines the advantage of scattering model and coherent decomposition.