Scattering Model Based Polarization Similarity Measurement And Its Application For Despeckling

Polarization synthetic aperture radar(SAR) technique is a kind of advanced remote sensing technology based on synthetic aperture radar technique. Polarization SAR technology not only inherited the synthetic aperture radar technology throughout the day, all-weather, the advantages of high resolution, and polarization SAR can obtain a variety of echo signal, providing more target scattering information. Polarized SAR is widely used in civil and military fields. But polarization speckle noises in SAR image not only make the quality of SAR image information is greatly reduced, but also make target recognition, detection and classification difficult. Many existing speckle noise suppression algorithms in noise reduction and not well maintain the polarization scattering properties of SAR image. Polarization SAR based on synthetic and real data statistical analysis, put forward two new methods of similarity measure the scattering characteristics: polarization similarity measure method based on three categories and polarization similarity measure method based on statistical distribution model. And its application to the quality of the polarization SAR speckle noise suppression algorithms study, the main work is as follows:(1) Based on the study of the polarization similarity measurement method proposed by Wu yuezhen, puts forward a new method to measure similarity between two polarization SAR, polarization similarity measure based on three categories. The measurement of polarization SAR pixels first polarization target decomposed based on Hybrid four-component(HPD) decomposition, will get a vector to establish three-dimensional spatial polarization characteristics. Divide the data into three different classes and find the similar set of methods for each class. Each class of similarity threshold is different. Applied the similarity measure method to the quality of the polarization SAR speckle noise suppression algorithms study. Experimental results show that the similarity measure method can accurately to find the similar sets of pixels.(2) Based on the analysis of the scattering vector of the polarization SAR data, puts forward a new method to measure similarity between two polarization SAR. The method is based on statistical distribution model. The measurement of polarization SAR pixels first polarization target decomposed based on hybrid four-component(HPD) decomposition, will get a vector to establish three-dimensional spatial polarization characteristics. Through the various components of the difference between the pixel analysis of the scattering vector, find it approximate to normal distribution. Based on the statistical and analysis data, obtain the standard deviation of each component, according to the nature of the normal distribution, find the threshold. The absolute value of the difference by the comparison with a threshold value to determine whether two similar pixels. Applied the polarization similarity measure method to Lee filter study. Through the test, found that the method can accurately and is simple to measure the similarity of two pixels. Based on NL-Lee filtering algorithm, put forward a new algorithm for polarization SAR coherent speckle suppression based on scattering statistical model and NL-Lee. The experiments on both real and synthetic polarization SAR images show that, the algorithm achieved a better filtering effect than pretest filter algorithm and NL- Lee filter algorithm. This method can maintain the scattering characteristics of polarization SAR images.