基于分形几何的遥感图像纹理分析方法及应用

METHODOLOGY AND APPLICATION OF REMOTE SEN SING IMAGE TEXTURE ANALYSIS BASED ON FRACTAL GEOMETRY

从遥感影像的空间结构特征入手，利用在图像分析中具有巨大发展潜力的分形学方法，引入表征图像灰度曲面和自然形状代表性模式来计算遥感影像窗口的分维值，通过数据转换形成可视化图形图像，并与常规纹理分析方法进行了综合对比。最后对上饶地区典型子区不同等级、不同层次的地质体单元及构造信息进行了具体分析，实现了岩体单元—超单元的详细划分；对地层的划分精度可以逐级达到组、段和亚段；对不易识别的微弱构造信息的增强与提取亦具有较好的效果，表现在对弱信息域的断裂构造的提取和对线性构造场分布信息的增强。

Based on the spatial structure features of remote sensing image ,the great potential fractal method in the field of image analysis is used and the corresponding mathematics model is quoted to calculate the fractal dimension value of remote sensing image window,and multidisciplinary analysis and comparison with conventional texture analysis method are made.Then,by means of data transformation, visual photographs and image are produced.Lastly geological body units and tectonic information of different levels in typical sub regions of Shangrao region are analyzed,and rock body units and super units are classified in detail.Strata are divided into groups,members and sub members The methods also can extract fracture information and enhance the linear tectonic field′s distribution information.