摘要
纹理是遥感图像的重要特征,它揭示了图像中辐射亮度值空间变化的重要信息。本文运用地统计学中的对数变差函数计算图像纹理,并与图像的光谱信息结合,进行图像岩性分类,分析了不同大小窗口纹理信息对分类精度的影响。结果表明,运用地统计学原理进行图像分类,可大大提高图像的分类精度;采用较大窗口提取的纹理信息参与分类能使总体分类精度提高,但某些岩性类的分类精度有所下降,建议在实际应用中,根据具体情况选择窗口的大小。
The texture is one of the important features of remote sensing images. In this paper, the image texture was extracted from Landsat TM data by means of semivariogram of logarithms, one of the geostatistic functions, and added to multispectral lithological classification. Different window sizes were used to extract textural information. The results of image classification show that the classification based on spectral data and geostatistical textural information can produce much higher overall accuracy than that based merely on spectral data. Moreover, for lithological discrimination based on multispectral data, the larger the window size for texture extraction is, the more accurate the classification result will become. In practice, however, other factors, such as the boundary effect and the accuracy of some important lithological units, need to be considered in choosing an appropriate window size.
出处
《国土资源遥感》
CSCD
2003年第3期45-49,共5页
Remote Sensing for Land & Resources
基金
教育部留学回国人员科研启动基金资助项目。
关键词
图像纹理
岩性分类
变差函数
对数变差函数
遥感应用
Geostatistical texture
Variogram
Variogram of logarithms
Lithological discrimination
