摘要
鉴于遥感影像分类结果具有空间异质性特征,直接应用现有的抽样方法对其进行精度评价存在样本信息冗余度高和评价过程耗时长等局限性,提出了一种利用空间异质性的遥感影像分类结果抽样精度评价方法。该方法包括:1)利用景观聚集度指数量化遥感影像的空间异质性,实现研究区域的空间区划;2)根据聚集度指数设定各区划空间的权重系数,选择用于精度评价的样本点;3)与参考数据比较,实现分类结果的精度评价。以城镇遥感影像分类结果的精度评价为例,将该方法与简单随机抽样、分层抽样和基于灰度共生矩阵的系统抽样方法进行对比。结果表明,该方法降低了精度评价样本点的信息冗余,提高了样本点的代表性,较适合于精度要求高的遥感影像分类结果精度评价。
The existing sampling methods are mostly based on the design of independent homogeneous units,which have the limitations of redundant information and time consumption when evaluating the accuracy of remote sensing image classification results.This paper proposes an accuracy assessment method for remote sensing classification results based on spatial heterogeneity,which includes three aspects:1)Quantify the spatial heterogeneity of remote sensing images based on landscape aggregation index,and then divide the research area into different spatial regions;2)Set the weighting coefficients of each spatial region,and select the sample point location for accuracy assessment;3)Compare the selected samples with corresponding reference data,and assess the accuracy of the classification results.Taking the accuracy assessment of urban remote sensing image classification results as an example,the proposed method is compared with the simple random sampling method,the stratified sampling method,and the systematic sampling method based on gray level co-occurrence matrix.Experimental results show that the proposed method shows greater performance with lower information redundancy and higher representativeness of selected samples,which is more suitable for the accuracy assessment of remote sensing classification results that require higher accuracy requirement.
作者
王振华
徐利智
纪晴
刘智翔
WANG Zhenhua;XU Lizhi;JI Qing;LIU Zhixiang(College of Information,Shanghai Ocean University,Shanghai 201306,China)
出处
《遥感信息》
CSCD
北大核心
2020年第6期25-31,共7页
Remote Sensing Information
基金
上海市地方院校能力建设项目(19050502100)
国家自然科学基金项目(41501419、41671431)。
关键词
精度评价
空间抽样方法
空间异质性
聚集度指数
遥感影像分类结果
accuracy assessment
spatial sampling
spatial heterogeneity
aggregation index
remote sensing classification result