Main-stream Feature Extraction of the Yellow River Based on Regional Spectral Un-mixing

摘要

The mainstream of the Yellow River is the water dynamic axis with maximum velocity, ferocity flow pattern and turbulence wave on water surface, which is an important basis for flood control decision and river regulation planning. The location of the main-stream is significant for situation forecasting and flood prevention. The development of the remote sensing technology makes it possible to extract the main-stream of the Yellow River quickly and exactly. Locating the position of the main-stream is a key to make decision for flood control. We propose a method of feature extraction for main-stream based on regional spectral un-mixing in this paper. Since the main-stream of the Yellow River is regarded as an endmember in high dimension space according to the spectral un-mixing technology, in which the geometric properties of the spectral vector in high dimensional space are utilized, the representative endmember of the main-stream area of the Yellow River could be selected thus the typical feature could be extracted accordingly. The experiment results on both synthesized and real data indicate that this method is effective on extracting the feature of main-stream of the Yellow River.