高分辨率遥感影像在土地利用分析中的精度验证  

Accuracy verification of high-resolution remote sensing images in land use analysis

作　　者：徐克成[1] XU Kecheng(Suzhou Surveying and Mapping Institute Co.,Ltd.,Suzhou 215008,China)

机构地区：[1]苏州市测绘院有限责任公司,江苏苏州215008

出　　处：《中国高新科技》2025年第3期152-154,共3页

摘　　要：文章基于苏州市某典型区域研究高分辨率遥感影像在土地利用分析中的精度验证。通过实地采样点验证,对比最大似然法、支持向量机、随机森林3种分类方法精度。结果表明,随机森林方法总体精度达89.6%,较其他两种方法分别高出5.3%、3.7%。研究建立了适合区域特征的精度评价体系,优化分类参数配置,提出边界区域精度提升策略。研究成果为提高土地利用遥感解译精度提供方法支撑,对完善精度验证体系具有参考价值。Based on a typical area in Suzhou city,the article studies the accuracy verification of high-resolution remote sensing images in land use analysis.Through field sampling point verification,three classification methods-maximum likelihood,support vector machine,and random forest-were compared for accuracy.The results show that the random forest method achieved an overall accuracy of 89.6%,which is 5.3%and 3.7%higher than the other two methods respectively.The study established an accuracy evaluation system suitable for regional characteristics,optimized classification parameter configuration,and proposed strategies for improving boundary region accuracy.The research results provide methodological support for improving the accuracy of land use remote sensing interpretation and have reference value for improving the accuracy verification system.

关 键 词：遥感影像 土地利用 精度验证 分类方法 精度评价 

分 类 号：P237[天文地球—摄影测量与遥感]