机载X波段极化干涉SAR数据非监督分类方法研究

针对经典极化分类算法在处理机载X波段SAR数据时将过多地物分为体散射类型,并且容易受噪声影响,分类结果存在大量误分现象的问题,通过对机载X波段SAR数据非监督分类方法的研究,提出将极化干涉信息用于机载X波段极化干涉SAR数据的分类。通过运用极化干涉数据进行目标分解得到参数A1和A2对数据进行初始分类,然后结合改进的Wishart最大似然分类算法来进行地物的自适应分类。实验结果表明,该方法能有效避免平地效应的影响,抗噪性好,能正确区分三种典型散射类型,分类效果明显优于极化分类效果。

Unsupervised Classification and Analysis for Airborne X-Band Polarimetric Interferometric SAR Data

An unsupervised classification approach is introduced for airborne X-band polarimetric interferometric SAR data sets,based on optimal coherence decomposition and an improved complex Wishart maximum-likelihood classifier.Traditional PolSAR unsupervised classification methods classify too many pixels into volume scattering category,and are apt to be affected by speckle noise,when applied to airborne X-band SAR data.So the polarimetric interferometric information is added in for unsupervised classification of airborne X-band SAR data on this question.First,the parameter A1 and A2 are figured out through target decomposition based on PolInSAR data.And then,the original data is initialized into 9 categories according to the A1-A2 plane.Last,the improved complex Wishart ML classifier is used for getting statistically self-adjusting class boundaries.The results show that the method performs better than traditional PolSAR and PolInSAR unsupervised classification methods distinctly with fine noise and flat earth effect immunity.