摘要
在遥感影像分类的过程中非光谱特征起着重要的作用。纹理特征作为一种重要的非光谱特征对于遥感影像分类精度的提高也有很重要的作用。以陇西黄土高原为实验区,Landsat TM5为数据源,利用灰度共生矩阵建立纹理特征统计量,通过实验分析不同地物提取过程中最有效的纹理特征量,并运用面向对象分类方法对其分类。结果表明,灰度共生矩阵提取的纹理特征对图像分类精度提高可起到一定的作用。
In this paper, taking the Loess Plateau of Longxi for the experimentation area, Landsat TM5 as the data source, we used the gray level co-occurrence matrix to set up the texture characteristics, and analyzed the most effective texture characteristic in the different feature extraction process by experiment. And then, we used the object-oriented method to classify these texture characteristics. The result shows that texture characteristics extracted by the gray co-occurrence matrix plays a role to the improved image classifi cation accuracy.
出处
《地理空间信息》
2015年第5期121-124,11,共4页
Geospatial Information
关键词
纹理特征
灰度共生矩阵
遥感
面向对象
texture characteristics,gray-level co-occurrence matrix,remote sensing,object-oriented
