基于BPLT模型的小麦叶片背景扣除方法的研究

为准确且无损测定小麦叶片的反射光谱,研究了不同背景对叶片表面反射光谱的影响,在400~1 000nm波段范围测定了小麦叶片8种背景下的反射光谱以及叶绿素含量。以PLATE模型为基础,首次提出了BPLT(background plate)模型,扣除由不同背景导致的叶片反射光谱的变化。模型以背景下叶片的反射率R_0,不同背景反射率σ为输入,空气和致密叶片的界面反射比R_(12),致密叶片和空气的界面反射比R_(21),致密叶片的透射系数τ三参数中间变量,得到最终无背景时叶片反射率R值的2-3-1模型。采用方差分析法(analysis of variance,ANOVA)进行了BPLT模型验证,对比分析了背景扣除前后10种叶绿素指数值的变化。结果表明,当反演的确定系数DC(determination coefficients)0.90且残差平方和SSE1时,反演的灵敏度较高,对小麦叶片不同叶绿素浓度的背景扣除有着较好的效果;采用BPLT模型背景扣除后,背景因素所占的百分比低于5%;优选了NDIMCARI的函数关系,NDIMCARI的斜率和叶绿素浓度的R~2由背景扣除前的0.847 4提高到背景扣除后的0.977 8。为真实测定不同背景下小麦叶片的反射光谱提供了依据。

Method of Background Elimination for Wheat Leaves Based on the BPLT Mo de l

In order to precisely acquire leaf reflectance spectra (400~1 000 nm) ,influence of background on leaf reflectance spectra was studied .Experiment was conducted to discriminate the characteristics of wheat leaves based on 8 background materi‐als and leaf chlorophyll concentration .BPLT (Background Plate) model ,based on the Plate model ,was promoted and applied to remove the influence of leave background .The BPLT model needed“2‐3‐1”variables ,which were input variables R0 (reflectance of the interaction of leaves and background ) ,σ(reflectance of background alone) ,intermediate variables R12 (reflectance of inter‐acting interface from air to a compact leaf ) ,R21 (reflectance of interacting interface from a compact leaf to air ) ,τ(the transmis‐sivity of the plate) ,and ultimate variable R (reflectance of a compact leaf alone) .To verify this model ,Analysis of Variance (ANOVA) was conducted to compare ten vegetation indices under different background influences .The results indicated that percent of variation in background reflectance associated with spectral vegetation indices was 5% lower after using the BPLT model .Meanwhile ,wheat leaf chlorophyll concentration at different levels could be effectively estimated by the means of BPLT model with determined coefficients (DC) greater than 0.9 and residual sum of squares (SSE) less than 1 .As with the ANOVA , vegetation indices NDI and MCARI were better than the other 8 ones .The slope of NDI&MCARI plotted as a function of mean wheat leaf chlorophyll concentration .R2 ranged from 0.847 8 to 0.977 8 with the applied method of BPLT model .The BPLT model is a powerful and accurate method for the acquisition of wheat leaf reflectance information .