Speckle Reduction, Based On Independent Component Analysis Of Remote Sensing Image

The polarimetric SAR system is an important area in remote sensing technology research and application now. Because of the advantage of obtaining multi-polarimetric images, polarimetric SAR is playing a more and more important role in gathering the detailed information regarding the geometric and electric structures of surfaces and covers of the earth. It has been proved that using polarimetric information can improve performance of SAR image processing and application. The speckle reduction which has been researched for a long time will produce important influence on the post-processing of polarimetric SAR images. Up to now, many scholars have brought forward lots of methods to reduce speckle. These methods impulse the further research on speckle. In this dissertation, a new polarimetric speckle reduction method is proposed by using Independent Component Analysis ( ICA ) which based on the multi-polarimetric trait of polarimetric SAR and the independence between texture SAR imagery and speckle. This is a new application of ICA technology in remote sensing area. The aim of ICA is to recover independent source signals from mixed signals when sources and mixed parameters are unknown which make ICA has high practical value and wide application. So this dissertation introduce ICA in the image processing of polarimetric SAR in order to separate texture image and speckle. The experiment shows ICA algorithm can separate speckle from polarimetric SAR image effectively and improve the performance of the image. Furthermore, the realization of ICA is simple and its calculation is convenience and efficiency. In conclusion, compared with the existing speckle reduction methods, ICA has its particular predominance on speckle reduction.