期刊文献+

纹理特征对森林蓄积量反演模型的影响 被引量:9

Texture features influences on inversion model of forest stock volume
下载PDF
导出
摘要 【目的】蓄积量是反映森林资源质量的重要指标,传统人工蓄积量调查方式费时耗力。遥感技术在林业中的应用能有效地弥补人工调查的缺陷,采用遥感技术进行森林蓄积量的反演是区域范围内蓄积量估测的一种重要手段。现有的遥感蓄积量估测方法中,对于纹理特征因子的选取没有得到足够的重视。随着高分辨遥感影像的不断涌现,影像纹理特征越来越明显,将纹理特征引入到森林蓄积量估测模型当中,是一个很好的尝试。但纹理特征是否有利于森林蓄积量的估测,以及如何影响森林蓄积量的估测,目前并不清楚。【方法】利用国产GF-1号为数据源,在数据预处理基础上,采用不同窗口大小提取的纹理信息,以及对纹理因子进行改进,研究其对于森林蓄积量反演模型精度的影响。【结果】1)改进纹理特征能有效提高蓄积量反演模型的精度。通过计算出遥感影像纹理均值改进指数、波段纹理均值改进植被指数和均值改进植被指数,结合地理因子,采用多元逐步回归方法构建森林蓄积量反演模型,结果精度有较大改善。2)纹理窗口大小为9×9时,森林蓄积量反演模型精度最高。提取3×3、5×5、7×7、9×9、11×11这5种窗口大小的纹理特征参数,分别构建森林蓄积量估测模型,并进行不同窗口下蓄积量反演精度进行检验。当窗口大小为9×9时模型效果最好,R^2最大,达到0.652,RMSE值最小(25.3545 m^3/hm^2),说明此时的窗口大小是最优模型窗口。【结论】当窗口为9×9时模型效果最好,此时的窗口大小是最优模型窗口。但是对于不同研究区在同一数据源下使用9×9窗口模型并不一定效果是最好的,纹理因子在建模中仍需根据研究区实际情况进行使用。 【Objective】It’s common and applicable in forestry remote sensing to use remote sensing information to establish model for estimating forest parameters.The estimation of forest growing stock has always been a difficulty in forestry remote sensing,of which model estimation is the common technique to be adopted.At present,there are many estimation models based on the relationship between forest stock and remote sensing factors.In these models,texture features are not paid enough attention.With the improvement of remote sensing spatial resolution,texture features become more and more obvious.It’s a good attempt to bring texture feature into forest stock estimation model to improve the estimation accuracy.However,now it is not clear whether the texture feature is beneficial to the estimation of forest stock and how it affects.【Method】This article uses Gaofen-1 satellite of China as the data source.On the basis of data preprocessing,different window sizes are adopted to extract texture information and the texture factor is improved to study its influence on the precision of forest stock inversion model.【Result】1)The improved texture features can effectively increase the accuracy of the forest stock inversion model.By computing improved index of texture mean,improved vegetation index of band texture mean and improved vegetation index of mean of remote sensing image,combining with geographical factors,and adopting multiple stepwise regression method to establish the forest stock inversion model,the accuracy of the results is improved greatly.2)When the texture window size is 9×9,the accuracy of the forest stock inversion model is highest.Five kinds of window size models were respectively extracted and calculated based on different windows,and the texture features were calculated respectively to establish forest stock estimation model,then accuracy tests are conducted for each model.When the window size is 9×9,the model effect is the best.R^2 is the largest reaching 0.652,and RMSE is the smallest,indicat
作者 叶子林 周普良 林辉 李新宇 YE Zilin;ZHOU Puliang;LIN Hui;LI Xinyu(Research Center of Forestry Remote Sensing&Information Engineering,Central South University&Technology,Changsha 410004,Hunan,China;Key Laboratory of Forestry Remote Sensing Based Big Data&Ecological Security for Hunan,Changsha 410004,Hunan,China;Key Laboratory of National Forestry and Grassland Administration on Forest Resources Management and Monitoring in Southern Area,Changsha 410004,Hunan,China)
出处 《中南林业科技大学学报》 CAS CSCD 北大核心 2020年第9期49-56,共8页 Journal of Central South University of Forestry & Technology
基金 “十三五”国家重点研发计划项目“人工林资源监测关键技术研究”(2017YFD0600900)。
关键词 林业遥感 蓄积量 多元逐步回归 纹理窗口 GF-1 forest remote sensing volume model multiple stepwise regression texture windows GF-1
  • 相关文献

参考文献18

二级参考文献157

共引文献216

同被引文献109

引证文献9

二级引证文献19

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部