本文以全极化SAR数据为研究对象。由于全极化数据相干矩阵T3或协方差矩阵C3服从复wishart分布,所以首先在此分布的基础上利用统计假设检验方法构建似然比参数,用以表征地表地物的变化程度,然后利用基于广义高斯分布模型的EM迭代算法(GGM-EM)对变化信息进行初提取,最后充分考虑上下文信息,利用概率松弛迭代算法对初检测信息进行优化。该方法不仅全自动提取变化信息,而且经过非相干平均、初始分类、分类结果优化3次降斑去噪处理,因此检测精度较高。通过与传统对数比值法的比较,证明该方法的有效性。

This paper takes fully polarimetric SAR data as study object to analyze the change detection technology. As the coherency matrix C3 or covariance matrix T3 of fully polarimetric SAR data follows a complex Wishart distribution. First, the likelihood-ratio parameter is built by a test statistic based on Wishart distribution to represent change features. Then the initial change information is extracted by the expectation maximization (EM) iterative algorithm based on a general Gaussian distribution. Finally, the change information is generated by optimizing the initial change information using probability relaxation iteration algorithm considering the context information. The method can extract change information automatically as well as produce change result of almost speckle noise free by integrating a series of filtering operations, including incoherent average, initial classifi cation, optimizing classifi cation. The validity of the method is demonstrated by comparison with traditional logarithm ratio method.