融合光谱和空间信息的黄河主溜特征提取

Main-Stream Feature Extraction of the Yellow River Fusing Spectral and Spatial Feature

本文基于多光谱遥感图像,提出了一种融合光谱特征和空间信息的主溜特征提取方法.将遥感影像的两大表现形式——光谱和空间信息结合,利用像元及其邻域的关系来描述其空间结构,用方向线来描述黄河主溜,从而确定其位置和走向.实验结果表明,该方法可有效地提取出黄河主溜特征.

A new mainstream feature extraction approach is proposed, which efficiently fuses the two main manifestations in remote sensing images： spectral features and spatial information. The approach depicts the spatial structures with the rdationships between pixels and their neighborhoods, and describes the main-stream with the direction line. Finally, the position and trend of the main-stream line are achieved. The experiments results show that the presented approach is competent for main-stream feature extraction and the feature could be extracted effectively.