人民黄河2025，Vol.47Issue(5)：51-57,7.DOI:10.3969/j.issn.1000-1379.2025.05.009
基于高分辨率影像的黄河内蒙古河道冰型分类方法研究
Classification Method of River Ice Types in Inner Mongolia Section of the Yellow River Based on High Resolution Images
摘要
Abstract
Juxtaposed ice,consolidated ice and open water are the main ice types during the frozen-up period alone the Inner Mongolia section of the Yellow River.The distribution and area of the ice types are important ice condition information that needs to be mastered for ice preven-tion decision-making.Taking the formation and characteristic analysis of river ice types in Inner Mongolia as the entry point,using support vec-tor machine and random forest as classifiers,and multi-band spectra,multi-band spectra+NDSI+NDFSI,and multi-band spectra+texture fea-tures as feature vectors,six experimental schemes were combined,and classification experiments and accuracy evaluations of river ice surface types were conducted based on the high-resolution remote sensing images of ZY-3.The results show that a)the supervised classification scheme using support vector machines as classifiers and multi-band spectra+NDSI+NDFSI as features has the highest overall accuracy for river ice type classification,especially improving the recognition ability of consolidated ice and open water.b)Comparing with the near-infrared single band threshold method,supervised classification with SVM and RF has higher accuracy in classifying river ice types in high-resolution images.This study has provided a method for automatic classification of river ice types in domestic high-resolution remote sensing images.关键词
河道冰型/高分辨率遥感影像/支持向量机/随机森林/黄河内蒙古段Key words
river ice type/high resolution image/support vector machine/random forest/Inner Mongolia reach of Yellow River分类
计算机与自动化引用本文复制引用
韩琳,何厚军,张秀伟,张宝森..基于高分辨率影像的黄河内蒙古河道冰型分类方法研究[J].人民黄河,2025,47(5):51-57,7.基金项目
陕西省语音与图像信息处理实验室开放基金资助项目(SASIIP-KFKT-2018-2) （SASIIP-KFKT-2018-2）
郑州师范学院博士启动资金项目(702344) （702344）
国家自然科学基金资助项目(U2243221) （U2243221）
