摘要
本文以英罗港为研究区,利用SPOT5假彩色合成影像,结合现场踏勘,提取了研究区内分布较广且面积较大的4种红树林种类(白骨壤、桐花、秋茄和红海榄)的图谱特征,并建立了相应的遥感解译标志。利用SVM方法开展了研究区内红树林的分类制图,采用样点验证方法评价提取精度,结果显示,对4种红树林的总体分类精度为86.67%,Kappa系数为0.82。
Taking the Yingluo Port as the study area,we analyzed the image features of four dominant species of man-grove （Avicennia marina,Aegiceras corniculatum,Kandelia candel and Rhizophora stylosa）in SPOT-5 false-color com-posite image,supported by the field survey.And the interpretation signs were established accordingly.The SVMmethod was adopted to extract mangrove species from satellite image,and the validation results showed that the overall accuracy was 86.67% and the Kappa coefficient was 0.82.
出处
《激光生物学报》
CAS
CSCD
2014年第6期609-613,共5页
Acta Laser Biology Sinica
基金
中欧国际合作龙计划项目(ID:10470)
海洋公益性行业科研专项经费项目(201205008-1)
关键词
遥感
红树林
图谱特征
remote sensing
mangroves
image features
