Polarimetric Sar Image Speckle Reduction

The polarimetric SAR is a new type imaging radar to measure polarimetric characteristic of radiant signals and combines both advantage of measuring full complex scattering matrix for each resolution cell in a scene and producing two-dimensional high resolution images, which greatly improve its ability to discern surface features. Polarimetric SAR is thus playing a more and more important role in research on remote sensing and application area. It is widely used in both civil and military areas now. Since SAR and polarimetric SAR is a coherent imaging system, its inherent speckle noise will produce important influence on the post-processing of SAR images, such as target detection, classification and recognition, etc. So, speckle reduction becomes to the first step of SAR images post-processing.Up to now, the speckle problem is still a pinch point in SAR images processing field. The classic spatial filtering algorithms and the refined algorithms can restrain speckle to a certain extent, but they do not make full use of polarimetric information when applied to speckle reduction for polarimetric SAR images. The dissertation studies the method of reducing speckle with independent component analysis after analyzing speckle characteristic in polarimetric SAR images. Independent Component Analysis (ICA) is a new multi-dimensional signal processing technology developed from Blind Source Separation of late years. Some scholars have brought forward a new polarimetric speckle reduction method by using Independent Component Analysis which based on the multi-polarimetric trait of polarimetric SAR and the independence between speckle and texture of SAR imagery. While the algorithm thinks that there is only one real scene in polarimetric SAR source images, so they get only one recovering image in the procedure of image separation and lose polarimetric property. In this dissertation, the method of sparse coding shrinkage based on independent component analysis is proposed to reduce speckle in polarimetric SAR images. The experiment results show that the method can reduce speckle effectively and improve the performance of the image evidently. Furthermore, the realization of ICA is simple and its calculation is convenience and efficiency.