基于SVR的GF1号遥感影像森林蓄积量估测

Estimation of Forest Volume of GF1 Remote Sensing Image Based on SVR

国产高分辨率卫星遥感影像已成为森林资源调查和监测的重要数据源。基于国产卫星遥感的森林蓄积量估算成为重要的研究方向之一。本研究以福建省将乐县为研究区,选择国产高分辨率高分一号卫星2 m分辨率遥感影像为主要数据源,加以辅助野外实地调查数据,分别采用多元线性回归和SVM(support vector machine)回归方法开展亚热带针叶林蓄积量估算效果评价研究。首先,从融合影像中提取遥感因子,包括11个光谱因子和10个纹理因子等;其次,对21个遥感因子进行相关性分析,选取皮尔森相关系数较大的6个遥感因子;第三,应用多元线性回归和支持向量机回归(support vector regression,SVR)对所选遥感因子建立模型,选取最优模型反演将乐县蓄积量分布图。结果表明:支持向量机回归(SVR)估测蓄积量的模型预估精度达到98.22%。

High resolution satellite remote sensing image data has become the main source of forest resource survey and monitoring. The domestic satellite remote sensing forest volume estimation has become one of the important research direction. The paper takes Jiangle County, Fujian Province as the research area, selects the High resolution GF -1 satellite with 2 m resolution remote sensing image made in China as the main data source, and assists the field investigation data. Multiple linear regression and support vector machine (SVM) regression were used to evaluate the effect of subtropical coniferous forest volume estimation. Firstly, a series of remote sensing factors such as 11 spectral factors and 10 texture factors were extracted from the fusion image. Secondly, the correlation of 21 remote sensing factors was analyzed and 6 remote sensing factors with large Pearson correlation coefficient were selected. The multiple linear regression and support vector regression (SVR) were used to establish the model of the selected remote sensing factors, and the optimal model was selected to retrieve the distribution map of the volume in Jiangle County. The results show that the prediction accuracy of SVR model is 98.22%.