测绘通报 ›› 2022, Vol. 0 ›› Issue (7): 138-142,153.doi: 10.13474/j.cnki.11-2246.2022.0218

• 技术交流 • 上一篇    下一篇

加权全极化SAR图像分类下的功能区土地时空变化特征提取

杨俊1,2, 刘灵辉3   

  1. 1. 湖南省第二测绘院, 湖南 长沙 410114;
    2. 自然资源部南方丘陵区自然资源监测监管重点实验室, 湖南 长沙 410114;
    3. 电子科技大学公共管理学院, 四川 成都 611731
  • 收稿日期:2021-12-10 修回日期:2022-05-20 出版日期:2022-07-25 发布日期:2022-07-28
  • 作者简介:杨俊(1982—),女,高级工程师,主要研究方向为国土空间规划、土地综合整治、生态保护修复、自然资源调查评价等。E-mail:daodaocheng95j@163.com
  • 基金资助:
    2020年度国家社会科学基金(20BGL228)

Feature extraction of spatial and temporal change of land in functional areas based on weighted full-polarization SAR image classification

YANG Jun1,2, LIU Linghui3   

  1. 1. The Second Surveying and Mapping Institute of Hunan Province, Changsha 410114, China;
    2. Key Laboratory of Natural Resources Monitoring and Supervision in the South hilly Area of the Ministry of Natural Resources, Changsha 410114, China;
    3. School of Public Affairs and Administration, University of Electronic Science and Technology of China, Chengdu 611731, China
  • Received:2021-12-10 Revised:2022-05-20 Online:2022-07-25 Published:2022-07-28

摘要: 为准确分析功能区土地时空演变情况,需准确区分土地时空变化的相关特征,本文设计了功能区土地时空变化特征提取模型。首先采用全极化分解和灰度共生矩阵,对SAR图像中反映功能区土地时空变化的、不同地物的各类散射特征和纹理特征进行分类。然后确定最佳加权全极化特征组合,将该组合输入随机森林模型,完成最终图像中地物分类。最后以湖南省某市生态功能区土地时空变化特征为例,实现功能区土地时空变化特征分类提取。测试结果表明,该模型采用加权全极化特征组合,可准确描述地物分布情况,保证地物的可靠分类,能实现较好的提取效果。

关键词: 加权全极化, SAR图像分类, 功能区土地, 时空变化, 特征提取, 特征组合

Abstract: In order to accurately analyze the spatial-temporal evolution of land in functional areas, it is necessary to accurately distinguish the relevant characteristics of land spatial-temporal change. This paper designs the feature extraction model of land spatial-temporal change in functional areas. By using full polarization decomposition and gray level co-occurrence matrix,all kinds of scattering features and texture features of different objects in SAR image which reflect the spatiotemporal change of land in functional area are classified, and the best weighted full polarization feature combination is determined.The combination is input into random forest model to complete the classification of ground objects in the final image and realize the feature extraction of spatiotemporal change of land in functional area. The test results show that: the model uses weighted full polarization feature combination,which can accurately describe the distribution of surface features and ensure the reliable classification of surface features.Taking the spatial-temporal change characteristics of land in an ecological function area of Hunan province as an example, it can achieve good extraction effect.

Key words: weighted total polarization, SAR image classification, functional area land, spatiotemporal variation, feature extraction, characteristics combination

中图分类号: