论文部分内容阅读
K-Wishart分布旨在通过统计方法更精确地描述极化SAR多视协方差矩阵或相干矩阵数据,揭示极化SAR影像在异质场景下的非高斯统计特性。以内蒙古自治区依根实验区和河北省遵化实验区的国内机载数据为例,分别进行了Wishart和K-Wishart非监督分类实验。研究结果表明,K-Wishart分类器适用于提取林地、园地、农村居民点等较不均匀区域。同时,本文通过分类准确性和稳定性两个方面对K-Wishart分类器的性能进行了评价。
The K-Wishart distribution aims to reveal the non-Gaussian statistical properties of polarized SAR images in heterogeneous scenes by using statistical methods to describe polarimetric SAR multi-covariance covariance matrix or coherence matrix data more accurately. Taking the domestic airborne data of Izu Experimental Zone of Inner Mongolia Autonomous Region and Zunhua Experimental Zone of Hebei Province as an example, Wishart and K-Wishart unsupervised classification experiments were carried out respectively. The results show that the K-Wishart classifier is suitable for the extraction of less uniform areas such as forest land, garden land and rural settlements. At the same time, this paper evaluates the performance of K-Wishart classifier through classification accuracy and stability.