摘要
为提高勘察丘陵山区森林覆盖面积的效率,降低劳动强度,本文利用无人机搭载可见光相机获取了威海双岛林场(区域1)和烟台栖霞(区域2)两处林地的可见光图像,并计算可见光图像的植被指数,结合像元二分法构建了11种森林植被覆盖度提取模型,提取了两个区域的植被覆盖度。将支持向量机分类结果作为真值,对11种模型提取结果进行验证,筛选出误差小于11%的提取模型。利用烟台栖霞获取的林地可见光图像作为测试集,依据两地植被覆盖度的精度结果进行提取模型的普适性评价。研究表明,在两个试验区域中,裸地(非植被)像元信息的误差置信度为4%时,在11种森林植被覆盖度提取模型中,COM模型提取的林地植被覆盖度精度最高,与真值的精度在区域1为91.87%、区域2为89.34%。在11种森林植被覆盖度提取模型中,COM模型在两地提取植被信息的误差最小,具有更好的普适性。研究结果可为丘陵山区森林覆盖率高效提取和林区荒漠化监测提供技术支持。
In order to improve the efficiency of investigating the forest cover area in hilly mountainous areas,and reduce the labor intensity,this paper used the Unmanned Aerial Vehicle(UAV)remote sensing platform to obtain the visible light images of two blocks of woodland in the Weihai Shuangdao Forest Farm(area 1)and the Yantai,Qixia(area 2).The vegetation index of visible light images was calculated.11 kinds of Fractional Vegetation Cover(FVC)extraction models were established by combining with the pixel dichotomy method.Then the Fractional Vegetation Cover(FVC)of the two areas was extracted.The classification results of Support Vector Machine(SVM)were used as the true values that verified the extraction results of 11 kinds of extraction models.The extraction models with an error less than 11%were selected in this paper.The visible images of forest land obtained from Qixia,Yantai were used as the test set.The generalizability of the extraction models was evaluated based on the accuracy results of the vegetation cover in the two places.The results showed that the accuracy of the vegetation cover extracted from the COM model was the highest among the 11 kinds of Fractional Vegetation Cover(FVC)extraction models in the two study areas,with the error confidence level of 4%for the bare land(non-vegetation)pixel information,and its accuracy with the true value was 91.87%in area 1 and 89.34%in area 2.Among the 11 kinds of Fractional Vegetation Cover(FVC)extraction models,the COM model had the least error in extracting vegetation information in both regions and had better generalizability.The results of the study could provide technical supports for efficient extraction of forest cover in hilly mountainous areas and desertification monitoring in forest areas.
作者
黎文华
兰玉彬
肖啸
苗建驰
刘瑶
田秉权
赵静
LI Wenhua;LAN Yubin;XIAO Xiao;MIAO Jianchi;LIU Yao;TIAN Bingquan;ZHAO Jing(School of Agricultural Engineering and Food Science,Shandong University of Technology,Zibo 255000,China;Shandong University of Technology Sub-center of National Center for International Collaboration Research on Precision Agricultural Aviation Pesticide Spraying Technology,Zibo 255000,China;Shandong Xiahe Green Prevention and Con⁃trol Research Institute Co.Ltd.,Jinan 250000,China)
出处
《内蒙古农业大学学报（自然科学版）》
CAS
2022年第6期40-50,共11页
Journal of Inner Mongolia Agricultural University(Natural Science Edition)
基金
山东省自然科学基金项目(ZR2021MD091)
山东省引进顶尖人才“一事一议”专项经费资助项目(鲁政办字[2018]27号)
“基于地空一体化多源遥感数据的松材线虫病识别与监测”项目
关键词
可见光图像
支持向量机
像元二分法
森林植被覆盖度提取模型
植被覆盖度
Visible light image
Support vector machine
Pixel dichotomy method
Fractional vegetation cover extraction models
Fractional vegetation cover
作者简介
黎文华(1994—),男,硕士研究生,主要从事遥感技术与路径导航提取方面的研究;通信作者:赵静,E-mail:zbceozj@163.com
