Method for Estimating Leaf Area Index of Winter Rapeseed Based on Fusion of Vegetation Indices and Texture Features

doi:10.12133/j.smartag.SA202507018

Abstract

Abstract:

[Objective] Leaf area index (LAI) is a vital agronomic parameter that reflects the structure of crop canopies, photosynthetic capacity, and population growth status. It holds significant importance for the precision cultivation and management of winter oilseed rape. Traditional methods for measuring LAI, such as destructive sampling or the use of costly instruments, are often constrained by low efficiency, high costs, and limited adaptability. In contrast, unmanned aerial vehicle (UAV) remote sensing technology offers a novel approach to rapid and non-destructive LAI monitoring due to its advantages in high resolution and flexibility. However, reliance solely on spectral vegetation indices (VIs) frequently results in saturation phenomena at elevated LAI levels, complicating accurate representation of complex canopy structures. Consequently, this study aims to investigate the integration of vegetation indices with texture features (TFs) using machine learning techniques to enhance the estimation accuracy of LAI throughout the entire growth cycle of winter oilseed rape. [Methods] The research was conducted at an experimental site in Gao'an city, Jiangxi province, where 81 plots were established with varying sowing dates, densities, and fertilization treatments. These plots encompassed four critical growth stages: seedling, bud elongation, flowering, and pod development. A DJI Phantom 4 RTK multispectral UAV was employed to acquire image data, while ground-truth LAI data were concurrently collected using an LAI-2200C plant canopy analyzer, resulting in a total of 324 valid samples. From the multispectral imagery obtained, eight vegetation indices, namely normalized difference vegetation index (NDVI), optimized soil adjusted vegetation index (OSAVI), normalized difference red edge (NDRE), enhanced vegetation index (EVI), green normalized difference vegetation index (GNDVI), simple ratio (SR), difference vegetation index (DVI), and canopy infrared reflectance estimate (CIRE), were derived alongside reflectance values from five spectral bands: blue, green, red, red-edge, and near-infrared. Additionally, 40 texture features were extracted based on the Gray-Level Co-occurrence Matrix. To select the ten most representative features with minimal redundancy among these variables, the minimum redundancy maximum relevance (mRMR) algorithm was utilized. Subsequently, three machine learning algorithms, multiple linear regression (MLR), extreme gradient boosting (XGBoost), and support vector machine regression (SVR), were applied to develop models for estimating LAI. To assess model generalizability and mitigate overfitting risks during evaluation processes, a 3-fold GroupKFold cross-validation approach was implemented to ensure that samples originating from the same plot remained intact between training and testing sets. The performance of each model was rigorously evaluated using several metrics, including the coefficient of determination (R²), root mean square error (RMSE), and mean absolute error (MAE). [Results and Discussions] The results indicated that the LAI of winter oilseed rape exhibited a dynamic trend throughout its growth cycle, characterized by being "low at the seedling stage, high at the bud elongation stage, decreasing at the flowering stage, and dropping again at the pod stage". The LAI values ranged from 0.90 to 6.39, with a uniform sample distribution observed within each developmental phase. Most vegetation indices and texture features demonstrated a highly significant correlation with LAI (P < 0.001), with the SR and near-infrared entropy (NIR-entropy) exhibiting the strongest correlation (r = 0.81). In terms of feature selection, vegetation indices such as SR, CIRE, and NDRE along with texture features like NIR-entropy and G-variance maintained high selection frequencies among the top ten mRMR-selected features. This indicated their stable contribution to model construction. Regarding model performance, the fused vegetation and texture features (VTFs) model outperformed all other models evaluated; specifically, the VTFs-SVR model achieved superior estimation accuracy across the entire growth cycle (R²=0.90, RMSE=0.38, MAE=0.27). When compared to models utilizing only vegetation indices or solely texture features, the fused model demonstrated particularly enhanced performance during high-coverage stages such as bud elongation, effectively addressing issues related to spectral saturation. Residual analysis further confirmed that the VTFs model exhibited a more concentrated residual distribution, indicating significantly greater prediction stability than single-feature models. [Conclusions] The fusion of vegetation indices and texture features extracted from UAV-based multispectral imagery, combined with machine learning modeling, particularly the SVR algorithm, enabled high-accuracy, non-destructive estimation of LAI throughout the entire growth cycle of winter oilseed rape. Texture features effectively supplemented canopy structural information, showing strong complementarity, especially during high-LAI stages where spectral data is prone to saturation. Through mRMR feature selection and group-based cross-validation, the model demonstrated good generalizability and practical application potential. This method could provide reliable technical support for monitoring winter oilseed rape growth status and precision agriculture management. Future research would further incorporate canopy structural parameters or multi-temporal features to enhance the model's estimation capability during stages dominated by non-leaf organs.

Key words: winter rapeseed, leaf area index, mRMR algorithm, nested cross-validation, feature fusion, machine learning

CLC Number:

S565.4

LIU Jie, GUO Jiaxin, ZHANG Jiahao, ZHANG Bingchao, XIONG Jie, CAO Jianpeng, WU Shangrong, DENG Yingbin, CHEN Guipeng. Method for Estimating Leaf Area Index of Winter Rapeseed Based on Fusion of Vegetation Indices and Texture Features[J]. Smart Agriculture, 2025, 7(6): 161-173.

Figures/Tables 14

Fig. 1

Table 1

Table 2

Table 3

Calculation methods of vegetation indices

变量名	计算公式
NDVI^［19］	$ρ N - ρ R ρ N + ρ R$	（1）
OSAVI^［20］	$1.16 ρ N + ρ R + 0.16 × (ρ N - ρ R)$	（2）
NDRE^［21］	$ρ N - ρ R E ρ N + ρ R E$	（3）
EVI^［22］	$2.5 × ρ N - ρ R ρ N + 6 × ρ R - 7.5 × ρ B + 1$	（4）
GNDVI^［19］	$ρ N - ρ G ρ N + ρ G$	（5）
SR^［23］	$ρ N ρ R$	（6）
DVI^［24］	$ρ N - ρ R$	（7）
CIRE^［25］	$ρ N - ρ R E ρ R E$	（8）

Table 3

Table 4

Fig. 2

Fig. 3

Table 5

Fig. 4

Table 6

Fig. 5

Fig. 6

Fig. 7

Table 7

References 30

[1]	马宇靖, 吴尚蓉, 杨鹏, 等. 油料作物产量遥感监测研究进展与挑战[J]. 智慧农业(中英文), 2023, 5(3): 1-16.
	MA Y J, WU S R, YANG P, et al. Research progress and challenges of oil crop yield monitoring by remote sensing[J]. Smart agriculture, 2023, 5(3): 1-16.
[2]	BOCK A D, BELMANS B, VANLANDUIT S, et al. A review on the leaf area index (LAI) in vertical greening systems[J]. Building and environment, 2023, 229: ID 109926.
[3]	李方一, 黄璜, 官春云, 等. 作物叶面积测量的研究进展[J]. 湖南农业大学学报(自然科学版), 2021, 47(3): 274-282.
	LI F Y, HUANG H, GUAN C Y, et al. Research progress of crop leaf area measurement[J]. Journal of Hunan agricultural university (natural sciences), 2021, 47(3): 274-282.
[4]	徐乐园, 毛克彪, 郭中华, 等. 卷积神经网络在农业遥感图像语义分割中的应用综述[J]. 农业展望, 2024, 20(2): 70-75.
	XU L Y, MAO K B, GUO Z H, et al. Summary of application of convolutional neural network in semantic segmentation of agricultural remote sensing images[J]. Agricultural outlook, 2024, 20(2): 70-75.
[5]	MANDAPATI R. Remote sensing leaf area index (LAI) data assimilation with crop model for yield predictions in rice[D]. Odisha: Centurion University of Technology and Management, 2024.
[6]	胡健波, 张健. 无人机遥感在生态学中的应用进展[J]. 生态学报, 2018, 38(1): 20-30.
	HU J B, ZHANG J. Advances in the application of UAV remote sensing in ecology[J]. Acta ecologica sinica, 2018, 38(1): 20-30.
[7]	LI Z P, CHEN Z, CHENG Q, et al. Deep learning models outperform generalized machine learning models in predicting winter wheat yield based on multispectral data from drones[J]. Drones, 2023, 7(8): ID 505.
[8]	ZHU H Y, LIN C Z, DONG Z H, et al. Early yield prediction of oilseed rape using UAV-based hyperspectral imaging combined with machine learning algorithms[J]. Agriculture, 2025, 15(10): ID 1100.
[9]	PENG Y, ZHU T E, LI Y C, et al. Remote prediction of yield based on LAI estimation in oilseed rape under different planting methods and nitrogen fertilizer applications[J]. Agricultural and forest meteorology, 2019, 271: 116-125.
[10]	LI X C, ZHANG Y J, LUO J H, et al. Quantification winter wheat LAI with HJ-1CCD image features over multiple growing seasons[J]. International journal of applied earth observation and geoinformation, 2016, 44: 104-112.
[11]	ZHOU C, GONG Y, FANG S H, et al. Combining spectral and wavelet texture features for unmanned aerial vehicles remote estimation of rice leaf area index[J]. Frontiers in plant science, 2022, 13: ID 957870.
[12]	DU R Q, LU J S, XIANG Y Z, et al. Estimation of winter canola growth parameter from UAV multi-angular spectral-texture information using stacking-based ensemble learning model[J]. Computers and electronics in agriculture, 2024, 222: ID 109074.
[13]	WEI C W, HUANG J F, MANSARAY L R, et al. Estimation and mapping of winter oilseed rape LAI from high spatial resolution satellite data based on a hybrid method[J]. Remote sensing, 2017, 9(5): ID 488.
[14]	郭立笑, 陈志超, 马彦鹏, 等. 基于无人机多光谱和多波段组合纹理的马铃薯LAI估算[J]. 光谱学与光谱分析, 2024, 44(12): 3443-3454.
	GUO L X, CHEN Z C, MA Y P, et al. Estimation of potato LAI based on multi-spectral and multi-band combined texture of UAV[J]. Spectroscopy and spectral analysis, 2024, 44(12): 3443-3454.
[15]	TANG Z J, LU J S, ABDELGHANY A E, et al. Winter oilseed rape LAI inversion via multi-source UAV fusion: A three-dimensional texture and machine learning approach[J]. Plants, 2025, 14(8): ID 1245.
[16]	WANG G S, LAURI F, HASSANI A HEL. Feature selection by mRMR method for heart disease diagnosis[J]. IEEE access, 2022, 10: 100786-100796.
[17]	ZHANG X Y, LIU C. Model averaging prediction by K-fold cross-validation[J]. Journal of econometrics, 2023, 235(1): 280-301.
[18]	王俊, 吴振伟, 姜海, 等. 基于随机森林及遥感植被指数的无人农场水稻产量预测研究[J]. 智能化农业装备学报(中英文), 2025(2): 97-104.
	WANG J, WU Z W, JIANG H, et al. Rice yield prediction of unmanned farm based on random forest and remote sensing vegetation index[J]. Journal of intelligent agricultural mechanization, 2025(2): 97-104.
[19]	MANGEWA L J, NDAKIDEMI P A, ALWARD R D, et al. Comparative assessment of UAV and Sentinel-2 NDVI and GNDVI for preliminary diagnosis of habitat conditions in burunge wildlife management area, Tanzania[J]. Earth, 2022, 3(3): 769-787.
[20]	BINTE MOSTAFIZ R, NOGUCHI R, AHAMED T. Agricultural land suitability assessment using satellite remote sensing-derived soil-vegetation indices[J]. Land, 2021, 10(2): ID 223.
[21]	JORGE J, VALLBÉ M, SOLER J A. Detection of irrigation inhomogeneities in an olive grove using the NDRE vegetation index obtained from UAV images[J]. European journal of remote sensing, 2019, 52(1): 169-177.
[22]	TARIQ S, NAWAZ H, UL-HAQ Z, et al. Investigating the relationship of aerosols with enhanced vegetation index and meteorological parameters over Pakistan[J]. Atmospheric pollution research, 2021, 12(6): ID 101080.
[23]	LOLLI L, JOHNSON A, MONACO M, et al. The percentage of mature height as a morphometric index of somatic growth: A formal scrutiny of conventional simple ratio scaling assumptions[J]. Pediatric exercise science, 2023, 35(2): 107-115.
[24]	HUSSAIN S, RAZA A, ABDO H G, et al. Relation of land surface temperature with different vegetation indices using multi-temporal remote sensing data in Sahiwal region, Pakistan[J]. Geoscience letters, 2023, 10(1): ID 33.
[25]	QIAO K, ZHU W Q, XIE Z Y. Application conditions and impact factors for various vegetation indices in constructing the LAI seasonal trajectory over different vegetation types[J]. Ecological indicators, 2020, 112: ID 106153.
[26]	HUMEAU-HEURTIER A. Color texture analysis: A survey[J]. IEEE access, 2022, 10: 107993-108003.
[27]	BUGATA P, DROTAR P. On some aspects of minimum redundancy maximum relevance feature selection[J]. Science China information sciences, 2019, 63(1): ID 112103.
[28]	BISCHL B, LANG M, KOTTHOFF L, et al. Mlr: Machine learning in R[J]. Journal of machine learning research, 2016, 17(170): 1-5.
[29]	NIAZKAR M, MENAPACE A, BRENTAN B, et al. Applications of XGBoost in water resources engineering: A systematic literature review (Dec 2018–May 2023)[J]. Environmental modelling & software, 2024, 174: ID 105971.
[30]	AWAD M, KHANNA R. Support vector regression[M]// Efficient Learning Machines. Berkeley, CA: Apress, 2015: 67-80.

波段名称	中心波长/nm	面板反射率系数
Blue	450	0.760 1
Green	560	0.750 2
Red	650	0.749 4
Red edge	730	0.759 7
Near infrard	840	0.753 2

模型	核心参数	参数说明与取值范围
SVR	核函数（Kernel）	默认使用RBF核函数
	惩罚参数（C）	对数均匀分布： 1e0~1e2
	Epsilon	［0.01， 0.05， 0.1， 0.2］
	Gamma	［'scale'， 'auto'］ + 对数均匀分布： 1e-4~1e-1
XGBoost	学习率（Learning_rate）	均匀分布： 0.01~0.21
	最大树深（Max_depth）	整数均匀分布： 3~6
	最小子权重（Min_child_weight）	整数均匀分布： 1~5
	列采样比例（Colsample_bytree）	［0.6， 0.7， 0.8， 0.9， 1.0］
	Gamma	均匀分布： 0~0.5
	L1正则化（Reg_alpha）	［0， 0.1， 0.5， 1］
	L2正则化（Reg_lambda）	［0.5， 1， 5］
MLR	无特殊超参数	使用普通最小二乘

冬油菜生育期	样本数	最小值	最大值	平均值	标准差	CV
苗期	81	0.9	3.23	2.41	0.44	0.18
蕾薹期	81	1.73	6.39	5.05	0.76	0.15
花期	81	3.15	5.33	4.37	0.43	0.10
角果期	81	1.97	4.40	2.93	0.49	0.17
全生育期	324	0.90	6.39	3.69	1.20	0.33

特征名称	VIs模型选择频率	TFs模型选择频率	VTFs模型选择频率
SR	1.00	—	1.00
CIRE	1.00	—	1.00
NDRE	1.00	—	1.00
B	1.00	—	—
GNDVI	1.00	—	—
NIR	1.00	—	1.00
R	1.00	—	—
NDVI	1.00	—	—
RE	1.00	—	—
OSAVI	0.80	—	—
NIR-entropy	—	1.00	1.00
G-variance	—	1.00	1.00
NIR-mean	—	1.00	—
R-contrast	—	1.00	0.60
B-correlation	—	1.00	0.60
G-dissimilarity	—	1.00	—
NIR-homogeneity	—	0.80	—
RE-correlation	—	0.60	—
G-correlation	—	0.60	0.40
R-entropy	—	0.60	0.60

生育期	特征类型	SVR			XGBOOST			MLR
生育期	特征类型	R ²	RMSE	MAE	R ²	RMSE	MAE	R ²	RMSE	MAE
苗期	VIs	0.66	0.26	0.20	0.49	0.31	0.23	0.63	0.27	0.21
	TFs	0.02	0.43	0.31	0.01	0.43	0.32	0.05	0.43	0.33
	VTFs	0.59	0.28	0.21	0.51	0.30	0.23	0.67	0.25	0.20
蕾薹期	VIs	0.19	0.68	0.51	0.21	0.67	0.49	0.33	0.62	0.47
	TFs	0.43	0.58	0.41	0.31	0.63	0.43	0.33	0.62	0.49
	VTFs	0.54	0.52	0.40	0.29	0.64	0.47	0.42	0.58	0.44
花期	VIs	0.33	0.35	0.27	0.42	0.32	0.26	0.29	0.48	0.31
	TFs	0.32	0.35	0.27	0.27	0.36	0.28	0.32	0.35	0.27
	VTFs	0.37	0.34	0.26	0.52	0.30	0.22	0.24	0.37	0.29
角果期	VIs	0.29	0.41	0.32	0.41	0.37	0.29	0.39	0.38	0.31
	TFs	0.27	0.51	0.39	0.30	0.41	0.30	0.26	0.59	0.43
	VTFs	0.49	0.34	0.26	0.48	0.35	0.27	0.46	0.36	0.28
全生育期	VIs	0.86	0.44	0.35	0.88	0.41	0.31	0.85	0.46	0.37
	TFs	0.85	0.45	0.32	0.83	0.48	0.35	0.83	0.48	0.36
	VTFs	0.90	0.39	0.29	0.87	0.45	0.31	0.83	0.51	0.39