详细信息
基于Worldview-2影像的林木冠幅提取与树高反演 被引量:9
Study on crown diameter extraction and tree height inversion based on worldview-2 data
文献类型:期刊文献
中文题名:基于Worldview-2影像的林木冠幅提取与树高反演
英文题名:Study on crown diameter extraction and tree height inversion based on worldview-2 data
作者:孙华[1,2] 鞠洪波[2] 张怀清[2] 凌成星[2]
第一作者:孙华
机构:[1]中南林业科技大学林业遥感信息工程研究中心;[2]中国林业科学研究院资源信息所
年份:2014
卷号:34
期号:10
起止页码:45-50
中文期刊名:中南林业科技大学学报
外文期刊名:Journal of Central South University of Forestry & Technology
收录:CSTPCD;;北大核心:【北大核心2011】;CSCD:【CSCD_E2013_2014】;
基金:国家"十二五"863课题:"数字化森林资源监测关键技术研究"(2012AA102001);林业公益性行业科研专项(201104028)
语种:中文
中文关键词:林业遥感;林木冠幅;杉木人工林;图像分割;均值漂移算法;非线性联立方程组
外文关键词:forest remote sensing;tree crown width;Chinese fir plantation;image segmentation;mean shift algorithm;non-linear simultaneous equations
分类号:S771.8
摘要:以湖南省攸县黄丰桥国有林场杉木人工林为例,探讨林木冠幅提取与树高反演方法研究。基于Worldview-2影像,采用均值漂移分割算法开展样地内杉木冠幅信息提取。通过设置不同分割尺度确定最佳的冠幅分割参数为hs=10,hr=6,M=20。对提取的冠幅边界进行平滑处理,利用平滑后的影像冠幅与实测树高,分别建立了冠幅树高曲线估计模型和非线性联立方程组反演模型。其中以树高作为哑变量,建立的影像冠幅树高非线性联立方程组模型的拟合效果最佳,模型决定系数R2为0.899,模型的变动系数(CV),平均百分标准误差(EMPSE)均在10%以内,是树高反演的一种有效手段。
By taking the Chinese fir plantation in Huangfengqiao Forest Farm in Youxian county, Hunan province as the tested objective, the methods of tree crown width extraction and tree height inversion were studied. Based on worldview-2 data, the tree crown width information of the plantation was extracted by adopting mean shift segmentation algorithm. After setting a series of Image segmentation scale, the optimal crown width segmentation parameters’ values were determined, they are hs=10, hr=6, M=20.The extracted crown width border was smoothened, and the estimation model of tree crown=height curve and the inversion model of nonlinear simultaneous equations were respectively established based on the smoothed image’s crown width and measured tree height. Of them, the model of image tree crown-high nonlinear simultaneous equations set by taking tree high as the dummy argument had the optimal fitting results, the model coefficient of determination R2=0.899, the variation coefficient of model CCV and the average percentage standard errors EMPSE both were less than 10%, so the last selected model is an effective means for tree height inversion of Chinese fir plantation.
参考文献:
正在载入数据...