异质性是复杂目标场景合成孔径雷达(sAR)图像纹理信息变化程度的度量,本文针对基于变差系数、基于算术-几何均值比和基于信息论三种异质性分析方法的原理进行综述,给出了具体的测量实现方法与步骤,并对比了三种测量方法的测量结果,最后本文给出了异质性分析与测量应用于具有复杂目标场景SAR图像分割上的实验仿真,证明了异质性分析与测量能够描述SAR图像场景的复杂程度,为SAR图像分割提供有效的辅助手段。
Scene heterogeneity, is a SAR image feature which reflects the textural information variation. Three heterogeneity measurements based on coefficient of variation, the ratio of the Arithmetic to the Geometric Mean and information theory respectively were theoretically investigated, and the detailed methods were provided. Finally, the performance comparison among the heterogenity was represented through the experiment simulation, and the samples of application in image segmentation were presented. The conclusion is that scene heterogeneity measurement can describe the degree of complexity for SAR imagery, which helps to post- processing of SAR image such as image segmentation.