加入多尺度图像纹理的岩性分类

利用遥感图像进行岩性分类,是遥感地质应用的重要方面之一.本文运用地统计学中变差函数提取图像纹理,并与原始的光谱图像相结合用于遥感图像的岩性分类.文章分析了不同尺度的纹理信息对岩性分类的作用,并进一步分析和比较多尺度图像纹理对岩性分类的作用.结果表明,在岩性分类过程中加入不同尺度的纹理信息可不同程度地提高图像的岩性分类精度,而同时加入多尺度的纹理信息,可得到更高的分类精度.将多尺度的图像纹理信息和光谱信息综合,是一种有效的岩性分类方法.

The use of multiscale texture in image classification for lithologic mapping

Lithologic mapping from image classification is one of the important aspects in geologic remote sensing applications. In this paper,the image texture features at different scales were extracted from Landsat TM image using the variogram as the texture measure,and then were combined with original multispeetral image in image classification for lithologic mapping.In order to evaluate the performance of texture features,texture features at single scale,and multiple scales were analyzed in terms of image classif...