梁 栋, 管青松, 黄文江, 黄林生, 杨贵军. 基于支持向量机回归的冬小麦叶面积指数遥感反演[J]. 农业工程学报, 2013, 29(7): 117-123.
    引用本文: 梁 栋, 管青松, 黄文江, 黄林生, 杨贵军. 基于支持向量机回归的冬小麦叶面积指数遥感反演[J]. 农业工程学报, 2013, 29(7): 117-123.
    Liang Dong, Guan Qingsong, Huang Wenjiang, Huang Linsheng, Yang Guijun. Remote sensing inversion of leaf area index based on support vector machine regression in winter wheat[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2013, 29(7): 117-123.
    Citation: Liang Dong, Guan Qingsong, Huang Wenjiang, Huang Linsheng, Yang Guijun. Remote sensing inversion of leaf area index based on support vector machine regression in winter wheat[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2013, 29(7): 117-123.

    基于支持向量机回归的冬小麦叶面积指数遥感反演

    Remote sensing inversion of leaf area index based on support vector machine regression in winter wheat

    • 摘要: 利用单一植被指数反演叶面积指数(LAI)时,存在不同程度的饱和性且每种指数只能包含部分波段的信息,该文提出利用支持向量机回归的方法进行叶面积指数的反演,可以用更多的波段信息作为输入参数以提高LAI反演精度。选取冬小麦起身期、拔节期和灌浆期的实测光谱和叶面积指数数据,用统计回归的方法分别建立NDVI-LAI和RVI-LAI模型,用支持向量机回归(SVR)方法分别建立以NDVI、RVI以及蓝、绿、红和近红外4个波段数据作为输入参数的回归预测模型,即NDVI-SVR、RVI-SVR和NRGB-SVR模型。上述5个模型分别利用对应时期的环境星HJ-CCD数据进行验证。结果表明:NDVI和RVI与叶面积指数(LAI)的回归模型预测的结果与实测值的RMSE分别为0.98与0.97;预测精度分别为59.2%与59.3%。以NDVI和RVI结合实测叶面积指数(LAI)训练并预测的结果与实测值的均方根误差RMSE分别为0.71与0.83预测精度分别为70.4%与67.1%。以蓝(B)、绿(G)、红(R)以及近红外(NIR)波段作为输入参数回归并预测的RMSE值为0.42,预测精度为81.7%。通过支持向量机回归预测具有更好的拟合效果,可以输入更多波段信息,提高了叶面积遥感反演精度,对冬小麦的多个生育期均具有较好的适用性。

       

      Abstract: Abstract: The method of inverting leaf area index (LAI) using a single vegetation index (VI) was influenced by different degrees of saturation and each index could contain in general two bands of information. This paper proposed the method of using support vector machine regression (SVR) for leaf area index inversion, which could use more band information as input parameters in order to improve LAI inversion accuracy. Using the winter wheat's actual spectra measurement and leaf area index data in the period of erecting stage, elongation stage and filling stage, we established a NDVI-LAI and RVI-LAI model with the statistical regression method respectively, and established regression prediction model using NDVI, RVI, as well as blue, green, red and near-infrared four-band data as input parameters with the support vector machine regression (SVR) method, namely the NDVI-SVR, RVI-SVR and NRGB-SVR model. The above five models used the corresponding period environment HJ-CCD data for validation respectively. The results showed that: the RMSE of 0.98, 0.97 with the prediction accuracy value of 59.2%, 59.3% was obtained using the NDVI-LAI and RVI-LAI regression model respectively, and the RMSE of 0.71, 0.83 with the prediction accuracy value of 70.4%, 67.1% was obtained using NDVI-SVR and RVI-SVR regression model respectively. With blue (B), green (G), red (R) and near infrared (NIR) bands as input parameters of support vector machine regression and prediction, the RMSE value is 0.39, the prediction accuracy value is 81.7%. Support vector machine regression (SVR) prediction has a better fitting effect, and can input more band information to improve the leaf area index remote sensing inversion accuracy which is suitable for winter wheat's multiple birth period.

       

    /

    返回文章
    返回