CAJ | 학술논문

以TM多光谱遥感影像为数据源,应用erdas软件提供的非监督分类、最大似然分类和子象元分类3种方法,对中国竹乡福建省顺昌县典型地物进行了分类和精度评价,研究结果表明,3种方法总体精度分别为:69.67%,78.00%,82.67%;Kappa系数分别为:63.6%,73.6%,79.2%.子像元分类法的总体精度和竹林识别精度均高于其他两种传统的分类方法,其竹林用户分类精度达80%,是进行竹林信息提取较为理想的方法.
이TM다광보요감영상위수거원,응용erdas연건제공적비감독분류、최대사연분류화자상원분류3충방법,대중국죽향복건성순창현전형지물진행료분류화정도평개,연구결과표명,3충방법총체정도분별위:69.67%,78.00%,82.67%;Kappa계수분별위:63.6%,73.6%,79.2%.자상원분류법적총체정도화죽림식별정도균고우기타량충전통적분류방법,기죽림용호분류정도체80%,시진행죽림신식제취교위이상적방법.
With TM multi-spectral remote sensing image as data source,three methods of unsupervised classification,maximum likelihood classification and sub-pixel classification provided by erdas software were used to classify and evaluate the typocal Chinese Bamboo County,Shunchang of Fujian Province.The results showed that the overall accuracy of three methods were 69.67%,78.00% and 82.67% respectively.Kappa coefficients were 63.6%,73.6%,79.2% respectively.The overall accuracy of sub-pixel classification and bamboo identification accuracy was higher than the other two traditional classification methods.The user classification accuracy of its bamboo forest was 80%,which was ideal for bamboo forest information extraction methods.