Page 84 - 《武汉大学学报（信息科学版）》2025年第9期

第 50 卷 第 9 期 武 汉 大 学 学 报（ 信 息 科 学 版 ） Vol.50 No.9
2025 年 9 月 Geomatics and Information Science of Wuhan University Sept. 2025

引文格式：郭浩然,张新,郑逸榛 . 基于样本迁移的无监督植被遥感制图方法[J]. 武汉大学学报（信息科学版）,2025,50(9):1812-
1824.DOI:10.13203/j.whugis20230242
Citation：GUO Haoran, ZHANG Xin, ZHENG Yizhen. Unsupervised Vegetation Remote Sensing Mapping Method Based on
Sample Migration[J]. Geomatics and Information Science of Wuhan University, 2025, 50(9): 1812-1824. DOI: 10.13203/j. whu⁃
gis20230242

基于样本迁移的无监督植被遥感制图方法

郭浩然 1,2 张 新 郑逸榛 1,2
1
1 中国科学院空天信息创新研究院遥感与数字地球全国重点实验室,北京，100101
2 中国科学院大学资源与环境学院,北京，100101

摘 要：长时序植被分类与动态制图对于地球表层生态环境变化及规律认知具有重要意义。针对长时序植被分类与动
态制图中存在的可靠样本稀缺、效率低、成本高的难题，提出一种基于历史植被分类图数据集自动获取可靠样本的无监
督分类方法，可实现对大标签样本数据的局部-全局样本优化迁移与分类制图。首先将历史大标签数据的几何属性与多
源遥感数据作为约束，实现大标签样本去噪以获得优化样本；然后通过长时序遥感影像变化检测实现可靠样本迁移；最
后使用随机森林方法对植被进行分层分类，获得长时序植被动态制图结果。以内蒙古自治区阿鲁科尔沁旗长时序植被
遥感制图为应用实例，实现了 2005−2022 年多时相遥感分类精度均优于 88%，验证了所提方法的科学性和实用性。
关键词：长时序遥感；样本优化；样本迁移；无监督分类；植被制图
中图分类号：P228 文献标识码：A 收稿日期：2024⁃08⁃12
DOI：10.13203/j.whugis20230242 文章编号：1671⁃8860（2025）09⁃1812⁃13
Unsupervised Vegetation Remote Sensing Mapping Method Based on
Sample Migration

1，2
GUO Haoran ZHANG Xin ZHENG Yizhen 1，2
1
1 State Key Laboratory of Remote Sensing and Digital Earth, Aerospace Information Research Institute, Chinese Academy of Sciences,
Beijing 100101, China
2 College of Resources and Environment, University of Chinese Academy of Sciences, Beijing 100101, China

Abstract： Objectives: Vegetation classification and mapping are of great significance to develop ecological
environmental protection. Supervised classification is the most widely used method for vegetation classifica⁃
tion and mapping because it can generate accurate classification results. However, most of the current vege⁃
tation classification and mapping methods rely on single-phase field data, and field sampling often requires
a lot of manpower and material resources. So it is difficult to realize long-time sequence dynamic vegetation
classification and mapping only by field sample data. Methods: This paper proposes an unsupervised classi⁃
fication method for automatically obtaining reliable samples based on historical vegetation classification map
datasets. Using the 1∶ 1 000 000 Chinese Vegetation Atlas dataset as prior knowledge, a model for opti⁃
mizing and migrating large-label samples for vegetation types is proposed. Multi-source data of the same or
similar temporal phase as the existing vegetation classification maps are used to construct a feature set for lo⁃
cal sample clustering optimization and global sample hierarchical Gaussian mixture optimization, which re⁃
places the manual selection of training samples, and then obtains a usable sample set. On this basis, the in⁃
variant region sample migration is carried out by combining the results of long-time series Landsat-derived
vegetation change analysis. The migrated samples are used for multi-temporal vegetation classification
mapping, to quickly obtain the dynamic mapping results of long-time series vegetation classification. Re⁃

基金项目：内蒙古科技重大专项（2021ZD0045）；国家重点研发计划（2021YFB3901301）。
第一作者：郭浩然，硕士，研究方向为植被生态遥感。guohaoran21@mails.ucas.ac.cn
通信作者：张新，博士，研究员。zhangxin@radi.ac.cn