Research On Speckle Filtering Of PolSAR Image

As an active microware, Polarimetric synthetic aperture radar(POLSAR) not only has the ability of cloud penetration and weather independence,but also can obtain the total polarimetric information of terrain. So it has become one of the focuses in the radar system and technology field. It has been widely used in military and civil areas,including military target detection and recognition, military combat effectiveness assessment, crop growth monitoring, disaster monitoring and assessment, hydrogeological monitoring, forest remote sensing and topographic mapping,etc. SAR and polarimetric SAR system is based on the principle of coherent imaging, so ite inherent speckle noise will make a great impact on the processing and imaging of SAR and polarimetric SAR images,such as reducing the radiation resolution,spatial resolution and gray level of the images, confusing image structures,etc. Therefore,reducing speckle is one of the most challenging problems in SAR images processing filed and becomes a very important and attractive research topic recently.Until to now,there are about two types of algorithms to suppress speckle noise.One is multilook processing before imaging,averaging of multiple looks during the generation of the SAR imgages.It suppress the speckle at the expense of spatial resolution of SAR and polarimetric SAR images.The other is the speckle filtering technique.It not only reduce speckle noise,but also preserve the spatial resolution of images. So speckle filtering methods are widely used in polarimetric SAR image processing. In the research of speckle reduction, a common problem is unable to maintain the texture features and scattering characteristics of pixels in SAR and polarimetric SAR image. The speckle filtering of the polarimetric SAR imgages is studied in this paper,and the main content and innovation are summarized as follows:Firstly,base on the further analysis of the speckle formation process, the speckle statistics, and the speckle models in the polarimetric SAR imgae,we have studied the classical speckle reduction algorithms and points out their advantages and disadvantages through the experimentsSecondly, on the basis of the above, we propose a new algorithm for speckle reduction algorithm based on the SMBSF filtering algorithm and using polarimetric interferometric SAR data. The new algorithm is named scattering-model-based speckle filtering for polarimetric interferometric SAR (Pol-InSAR) data.Thirdly,we apply the above proposed new filtering algorthm to classify polarimetric SAR images and also analyze the impact of filtering on the characteristic parameters representing the scattering properties. At the same time, we quantitatively assess the performance of the filter by introducing variation function and connectivity function for the structure analysis of spatial variability and connectivity analysis separately.Finally, we carry out the large number of experiments using true polarimetric SAR data. And the results show that the new speckle filtering method can more effectively suppress the speckle noise without blurring and preserve the scattering properties of pixels and the edge texture characteristic.Comparing to classical filtering algorithm the proposed algorithm is more advantageous.