遥感图像纹理分析方法

TEXTURAL ANALYSIS FOR REMOTELY SENSED IMAGERY

本文针对遥感图像纹理特征，定义纹理是“图像像元亮度值空间变化率的一种量度”。据此 定义，提出纹理强度和纹理密度两个新参量，定量地描述遥感图像的纹理特性。这两个参量反 映了图像中相邻象元亮度值的差别及图象中象元亮度值变化的频数。所设计的相应提取方法包 括：强度和法、绝对差法与最大、最小值法等三种纹理强度提取方法，以及两种纹理密度提取 方法：一阶纹理密度法与二阶纹理密度法。同时，讨论了局部移动窗口大小的选择原则。 本文不仅设计了从遥感图像亮度值的绝对变化提取纹理特征的方法，并进而提出相对梯度 概念，从象元亮度值的相对变化描述图象的纹理特征；由相对梯度法生成的纹理图像，不仅可 突出遥感图像的空间信息特征，并保留了原图像的部份波谱信息，原图像中的甚亮区和暗区、 甚暗区（如山地背阴坡）的纹理结构得以清晰地显现，效果独特。

According to defined the texture as the measurement of image gradient of grey, two parameters can be obtained to describe remote sensing image quantitatively. They are tile texture intensity and the texture density. The texture intensity is used to show differences of one pixel with its neighbours, while the texture density is used to show the frequency of changes of grey level. In addition the relative gradient are defined to describe the texture characteristic of image from tile relative changes of grey level. The producted texture images from this method include both the information of space and part of information of spectrum. So the texture structures in the overbright and over-dark ranges can be shown obviously. In this paper, we also discuss how to select the size of the moving window.