基于遥感数据随机模型的空间结构分析与蚀变信息提取

A Study of Spatial Structure Analysis and Alteration Information Extraction Based on Random Models of Remote Sensing Data

遥感图像信号属于随机信号,其波段数据直方图与二维散点图分别是随机模型一维、二维概率密度函数的基本估计。从遥感数据的随机模型出发,讨论了一维概率密度的基本类型与形态、二维高斯分布的椭圆几何参数特征等。在此基础上,系统分析了由不同类型直方图生成的二维散点图的空间几何结构特征,以及异常信息空间定位等问题。最后,通过应用实例阐述了遥感数据空间结构分析在蚀变信息提取中的重要性与实用性。

Remote sensing image signals are random signals whose band data histograms and two - dimensional scatter plots are respectively the basic estimates of one - dimensional probability density function and two - dimensional probability density function of the random model. In this paper, the basic types and patterns of one - dimensional probability density and the geometric parameter characteristics of Gaussian distribution ellipses are discussed based on the random models of remote sensing data. Then, problems such as the spatial geometric structure characteristics of the two - dimensional scatter plots generated by different histograms and the positioning in abnormal information space are analyzed. At last, the importance and practicability of the spatial structure analysis of remote Sensing data in alteration information extraction are demonstrated by a case study.