Estimation of Cotton Yield Based on Unmanned Aerial Vehicle RGB Images

doi:10.13304/j.nykjdb.2021.0918

Abstract

Abstract:

At present， unmanned aerial vehicle （UAV） system has been applied to crop yield estimation， the RGB camera carried by UAV was used to collect cotton canopy images from 3 heights （10， 20 and 30 m） at both flowering and boll development stage and boll-opening stage. The color and texture features were extracted from the images and integrated， and then the stepwise regression analysis and factor analysis were respectively performed to screen the most important features and establish cotton yield estimation models. By comparing the yield estimation models at different height and growth stages， the most appropriate growth stage and image acquisition height were determined. The results showed that the fitting degrees and precisions of the yield model at the flowering and boll development stage were better than those at the boll-opening stage at 20 and 30 m. However， the fitting degrees of models at two growth periods were relatively close at the height of 40 m， but the validation of the model was not significant at the boll-opening stage. By comparing the yield estimation models at the flowering and boll development stage at 20 and 30 m and the boll-opening stage at 40 m， it was found that the model established at 30 m at the flowering and boll stage by the stepwise regression analysis had the best fitting effect. Thus， it indicated that the flowering and boll development stage was the optimal growth period for cotton yield estimation， and 30 m was the best height for image collection. In summary， this paper showed that UAV RGB image could estimate cotton yield accurately and quickly， which provided theoretical and technical foundation for the estimation of cotton yield based on visible image， and provided reference for other crops to establish the yield estimation model.

Key words: UAV, image feature, stepwise regression analysis, factor analysis, cotton yield model

摘要：

目前，无人机系统已应用于作物产量估算，利用无人机搭载的RGB相机在花铃期和吐絮期从3个高度（10、20和30 m）分别采集棉花冠层图像，提取图像的颜色指数和纹理特征，进而对提取的特征分别进行逐步回归分析和因子分析，筛选出重要特征并构建棉花产量估算模型。通过对比分析2个生育时期和3个高度的产量估算模型，最终确定利用RGB图像对棉花进行产量估算的最佳生育时期和最佳高度。结果表明， 20 和30 m高度下花铃期图像建立的产量模型拟合度以及模型精度均比吐絮期好，而40 m高度下2个生育时期的模型拟合度接近，但花铃期的验证结果不显著；对比20和30 m高度下花铃期以及40 m高度下吐絮期的产量估算模型发现，30 m高度下花铃期通过SWR方法建立的模型拟合效果最佳，由此表明，棉花产量估算的最佳生育时期为花铃期，图像采集的最佳高度为30 m。综上，利用无人机RGB图像能准确快速估算棉花产量，为基于可见光图像的棉花产量估算提供了理论和技术参考，并为其他农作物估产模型的建立提供借鉴。

关键词: 无人机, 图像特征, 逐步回归, 因子分析, 棉花产量模型

CLC Number:

S126

Jing ZHANG, Simeng GUO, Yingchun HAN, Yaping LEI, Fangfang XING, Wenli DU, Yabing LI, Lu FENG. Estimation of Cotton Yield Based on Unmanned Aerial Vehicle RGB Images[J]. Journal of Agricultural Science and Technology, 2022, 24(11): 112-120.

张静, 郭思梦, 韩迎春, 雷亚平, 邢芳芳, 杜文丽, 李亚兵, 冯璐. 基于无人机RGB图像的棉花产量估算[J]. 中国农业科技导报, 2022, 24(11): 112-120.

Figures/Tables 6

References 39

1	张美娜,冯爱晶,周建峰,等.基于无人机采集的视觉与光谱图像预测棉花产量[J].农业工程学报,2019,35(5):91-98.
	ZHANG M N, FENG A J, ZHOU J F, et al.. Cotton yield prediction using remote visual and spectral images captured by UAV system [J]. Trans. Chin. Soc. Agric. Eng., 2019,35(5):91-98.
2	XU W, LAN Y, LI Y, et al.. Classification method of cultivated land based on UAV visible light remote sensing [J]. Int. J. Agric. Biol. Eng., 2019, 12(3): 103-109.
3	XU W, CHEN P, ZHAN Y, et al.. Cotton yield estimation model based on machine learning using time series UAV remote sensing data [J/OL]. Int. J. Appl. Earth Obs., 2021, 104: 102511[2021-10-27]. .
4	COLOMINA I, MOLINA P. Unmanned aerial systems for photogrammetry and remote sensing: a review [J]. ISPRS J. Photogram. Rem. Sens., 2014, 92: 79-97.
5	陶惠林,徐良骥,冯海宽,等.基于无人机数码影像的冬小麦株高和生物量估算[J].农业工程学报,2019,35(19):107-116.
	TAO H L, XU L J, FENG H K, et al.. Estimation of plant height and biomass of winter wheat based on UAV digital image [J]. Trans. Chin. Soc. Agric. Eng., 2019,35(19):107-116.
6	万亮,岑海燕,朱姜蓬,等.基于纹理特征与植被指数融合的水稻含水量无人机遥感监测[J].智慧农业,2020,2(1):58-67.
	WAN L, CEN H Y, ZHU J P, et al.. Using fusion of texture features and vegetation indices from water concentration in rice crop to UAV remote sensing monitor [J]. Smart Agric.,2020,2(1):58-67.
7	MUANGPRAKHON R, KAEWPLANG S. Estimation of paddy rice plant height using UAV remote sensing [J]. Eng. Access., 2021, 7(2): 93-97.
8	邓江,谷海斌,王泽,等.基于无人机遥感的棉花主要生育时期地上生物量估算及验证[J].干旱地区农业研究,2019,37(5):55-61, 69.
	DENG J, GU H B, WANG Z, et al.. Estimation and validation of above-ground biomass of cotton during main growth period using unmanned aerial vehicle (UAV) [J]. Agric. Res. Arid Areas, 2019,37(5):55-61, 69.
9	DU X, WAN L, CEN H, et al.. Multi-temporal monitoring of leaf area index of rice under different nitrogen treatments using UAV images [J]. Int J. Precis Agric. Aviat., 2020, 3(1):7-12.
10	CÓRCOLES J I, ORTEGA J F, HERNÁNDEZ D, et al.. Estimation of leaf area index in onion (Allium cepa L.) using an unmanned aerial vehicle [J]. Biosyst. Eng., 2013, 115(1): 31-42.
11	王来刚,徐建华,贺佳,等.基于无人机遥感的玉米叶面积指数与产量估算[J].玉米科学,2020,28(6):88-93.
	WANG L G, XU J H, HE J, et al.. Estimating leaf area index and yield of maize based on remote sensing by unmanned aerial vehicle [J]. J. Maize Sci., 2020,28(6):88-93.
12	申洋洋,陈志超,胡昊,等.基于无人机多时相遥感影像的冬小麦产量估算[J].麦类作物学报,2021(10): 1298-1306.
	SHEN Y Y, CHEN Z C, HU H, et al.. Estimation of water wheat yield based on UAV multi-temporal remote sensing image [J]. J. Triticeae Crops, 2021(10): 1298-1306.
13	汤勃,孔建益,伍世虔.机器视觉表面缺陷检测综述[J].中国图象图形学报,2017,22(12):1640-1663.
	TANG B, KONG J Y, WU S Q. Review of surface defect detection based on machine vision [J]. J. Image. Graph., 2017,22(12):1640-1663.
14	周元琦,王敦亮,陈晨,等.基于无人机RGB图像颜色及纹理特征指数的小麦产量预测[J].扬州大学学报(农业与生命科学版),2021,42(3):110-116.
	ZHOU Y Q, WANG D L, CHEN C, et al.. Prediction of wheat yield based on color index and texture feature index of unmanned aerial vehicle RGB images [J]. J. Yangzhou Univ. (Agric. Life Sci.), 2021,42(3):110-116.
15	刘欣谊,仲晓春,陈晨,等.利用无人机图像颜色与纹理特征数据在小麦生育前期对产量进行预测[J].麦类作物学报,2020,40(8):1002-1007.
	LIU X Y, ZHONG X C, CHEN C, et al.. Prediction of wheat yield using color and texture feature data of UAV image at early growth stage [J]. J. Triticeae Crops, 2020,40(8):1002-1007.
16	兰铭,费帅鹏,禹小龙,等.多光谱与热红外数据融合在冬小麦产量估测中的应用[J].麦类作物学报,2021,41(12): 1564-1572.
	LAN M, FEI S P, YU X L, et al.. Application of multispectral and thermal infrared data fusion in estimation of winter wheat yield [J]. J. Triticeae Crops, 2021,41(12): 1564-1572.
17	李昂. 基于无人机数码影像的水稻产量估测研究[D].沈阳:沈阳农业大学, 2018.
	LI A. Research on rice yield estimation based on digital imaging of UAV [D]. Shenyang: Shenyang Agricultural University, Master Dissertation, 2018.
18	胡莉婷,潘学标,王雪姣,等.基于COSIM模型的棉花区域产量动态预测研究——以新疆阿克苏市为例[J].棉花学报,2018,30(1):83-91.
	HU L T, PAN X B, WANG X J, et al.. Dynamic prediction of cotton regional yield based on the COSIM model-A case study of Akesu city, Xinjiang [J]. Cotton Sci., 2018,30(1):83-91.
19	王雪姣,潘学标,王森,等.基于COSIM模型的新疆棉花产量动态预报方法[J].农业工程学报,2017,33(8):160-165.
	WANG X J, PAN X B, WANG S, et al.. Dynamic prediction method for cotton yield based on COSIM model in Xinjiang [J]. Trans. Chin. Soc. Agric. Eng., 2017,33(8):160-165.
20	刘焕军,康苒, SUSAN U,等.基于时间序列高光谱遥感影像的田块尺度作物产量预测[J].光谱学与光谱分析,2016,36(8):2585-2589.
	LIU H J, KANG R, SUSAN U, et al.. Study on the prediction of cotton yield within field scale with time series hyperspectral imagery [J]. Spectrosc. Spectral Anal., 2016,36(8):2585-2589.
21	毛文华,王一鸣,张小超,等.基于机器视觉的田间杂草识别技术研究进展[J].农业工程学报,2004(5):43-46.
	MAO W H, WANG Y M, ZHANG X C, et al.. Research advances of weed identification technology using machine vision [J]. Trans. Chin. Soc. Agric. Eng., 2004(5):43-46.
22	LI S, YUAN F, ATA-UI-KARIM S T, et al.. Combining color indices and textures of UAV-based digital imagery for rice LAI estimation [J/OL]. Remote Sens., 2019, 11(15): 1763 [2021-10-27]. .
23	MAO W, WANG Y, WANG Y. Real-time detection of between-row weeds using machine vision [C/OL]// Proceedings of ASAE Annual Meeting. American Society of Agricultural and Biological Engineers, 2003:031004 [2021-10-27]. .
24	MEYER G E, NETO J C. Verification of color vegetation indices for automated crop imaging applications [J]. Comput. Electron. agric., 2008, 63(2): 282-293.
25	BENDIG J, YU K, AASEN H, et al.. Combining UAV-based plant height from crop surface models, visible, and near infrared vegetation indices for biomass monitoring in barley [J]. Int. J. Appl. Earth Obs., 2015, 39: 79-87.
26	LOUHAICHI M, BORMAN M M, JOHNSON D E. Spatially located platform and aerial photography for documentation of grazing impacts on wheat [J]. Geocart. Int., 2001, 16(1): 65-70.
27	唐银凤,黄志明,黄荣娟,等.基于多特征提取和SVM分类器的纹理图像分类[J].计算机应用与软件,2011,28(6):22-25, 46.
	TANG Y F, HUANG Z M, HUANG R J, et al.. Texture image classification based on multi-feature extraction and SVM classifier [J]. Comput. Appl. Software, 2011,28(6):22-25, 46.
28	TAMURA H, MORI S, YAMAWAKI T. Textural features corresponding to visual perception [J]. IEEE Trans. Syst. Man. Cybern., 1978, 8(6): 460-473.
29	王顺杰,齐春,程玉胜.Tamura纹理特征在水下目标分类中的应用[J].应用声学,2012,31(2):135-139.
	WANG S J, QI C, CHENG Y S. Application of Tamura texture feature to classify underwater targets [J]. J. Appl. Acoust., 2012,31(2):135-139.
30	周鹏飞,卢泽雨.基于SPSS多元线性回归模型在城市用水量的预测[J].水利科技与经济,2018,24(5):6-10.
	ZHOU P F, LU Z Y. Prediction of urban water consumption based on SPSS multiple linear regression model [J]. Water Conservancy Sci. Technol. Econ., 2018,24(5):6-10.
31	李凤.因子分析的原理与应用[J].产业与科技论坛,2014,13(10):76-77.
	LI F. Principle and application of factor analysis [J]. Ind. Sci. Tribune, 2014,13(10):76-77.
32	龚红菊.基于分形理论及图像纹理分析的水稻产量预测方法研究[D].南京:南京农业大学, 2008.
	GONG H J. Estimating paddy yield based on fractal and image texture analysis. [D]. Nanjing: Nanjing Agricultural University, 2008.
33	刘焕军,孟令华,张新乐,等.基于时间序列Landsat影像的棉花估产模型[J].农业工程学报,2015,31(17):215-220.
	LIU H J, MENG L H, ZHANG X L, et al.. Estimation model of cotton yield with time series Landsat images [J]. Trans. Chin. Soc. Agric. Eng., 2015,31(17):215-220.
34	BENDIG J, BOLTEN A, BENNERTZ S, et al.. Estimating biomass of barley using crop surface models (CSMs) derived from UAV-based RGB imaging [J]. Remote Sens., 2014, 6(11): 10395-10412.
35	毛树春,李亚兵,董合忠,等.中国棉花栽培学[M].第四版.上海:上海科学技术出版社,2019:405-410.
36	徐永胜,刘浩然,范伟伟,等.基于分水岭算法的无人机不同飞行高度下林木冠幅提取研究[J].西北林学院学报,2021,36(3):197-202, 259.
	XU Y S, LIU H R, FAN W W, et al.. Extraction of forest stand crown width at different flying heights of UAV based on watershed Algorithm [J]. J. Northwest Forestry Univ.,2021,36(3):197-202, 259.
37	张佩,陈郑盟,马顺登,等.用冠层叶色偏态分布模式RGB模型预测大豆产量[J].农业工程学报,2021,37(9):120-126.
	ZHANG P, CHEN Z M, MA S D, et al.. Prediction of soybean yield by using RGB model with skew distribution pattern of canopy leaf color [J]. Trans. Chin. Soc. Agric. Eng., 2021,37(9):120-126.
38	唐晏. 基于无人机采集图像的植被识别方法研究[D].成都:成都理工大学,2014.
	TANG Y. Research on the vegetation identification method based on UAV image acquisition [D]. Chengdu: Chengdu University of Technology, 2014.
39	郑小南,杨凡,李富忠.农作物图像分割算法综述[J].现代计算机,2020(19):72-75.
	ZHENG X N, YANG F, LI F Z. Overview of crop image segmentation algorithm [J]. Mod. Comput., 2020(19):72-75.

生育时期 Growth stage	因子Factor	20 m	30 m	40 m
花铃期 Flowering and boll stage	FA₁	-271.4	235.7	-180.7
	FA₂	-139.2	-197.1	-39.1
	FA₃	27.9	-69.2	158.8
吐絮期 Boll-opening stage	FA₁	224.8	204.5	160.7
	FA₂	-112.3	-110.0	63.5
	FA₃	104.3	100.5	165.2

生育时期 Growth stage	因子Factor	20 m	30 m	40 m
花铃期 Flowering and boll stage	FA₁	-271.4	235.7	-180.7
	FA₂	-139.2	-197.1	-39.1
	FA₃	27.9	-69.2	158.8
吐絮期 Boll-opening stage	FA₁	224.8	204.5	160.7
	FA₂	-112.3	-110.0	63.5
	FA₃	104.3	100.5	165.2

高度 Height/m	生育时期 Growth stage	分析方法 Analysis method	拟合模型 Fitting model	决定系数R²	RMSE/（kg·hm^-2）	NRMSE/%
20	花铃期 Flowering and boll stage	逐步回归SWR	Y=-3 598W+0.886 8D-1 250	0.733 3	184.4	5.71
	花铃期 Flowering and boll stage	因子分析FN	Y=-271.4FA₁-139.2FA₂+27.89FA₃+3 228	0.735 8	183.5	5.68
	吐絮期 Boll-opening stage	逐步回归SWR	Y=18 792I+1.952D+2 260	0.606 5	224.0	6.94
	吐絮期 Boll-opening stage	因子分析FN	Y=224.8FA₁-112.3FA₂+104.3FA₃+3 252	0.592 6	230.6	7.14
30	花铃期 Flowering and boll stage	逐步回归SWR	Y=-3 085W-3 102E+15 016	0.784 9	165.6	5.13
	花铃期 Flowering and boll stage	因子分析FN	Y=235.2FA₁-134.1FA₂-152.6FA₃+3 228	0.757 4	175.9	5.45
	吐絮期 Boll-opening stage	逐步回归SWR	Y=-296 8W+1 030	0.415 9	272.9	8.45
	吐絮期 Boll-opening stage	FN	Y=204.5FA₁-110FA₂+100.5FA₃	0.502 0	252.0	7.81
40	花铃期 Flowering and boll stage	逐步回归SWR	Y=-238 87M+11 146	0.435 5	268.3	8.31
	花铃期 Flowering and boll stage	因子分析FN	Y=-180.7FA₁-39.06FA₂+158.8FA₃+3 228	0.465 6	261.0	8.09
	吐絮期 Boll-opening stage	逐步回归SWR	Y=-2 369C+19 032	0.506 0	251.0	7.78
	吐絮期 Boll-opening stage	因子分析FN	Y=160.7FA₁+63.53FA₂+165.2FA₃	0.448 1	265.3	8.22

高度 Height/m	生育时期 Growth stage	分析方法 Analysis method	拟合模型 Fitting model	决定系数R²	RMSE/（kg·hm^-2）	NRMSE/%
20	花铃期 Flowering and boll stage	逐步回归SWR	Y=-3 598W+0.886 8D-1 250	0.733 3	184.4	5.71
	花铃期 Flowering and boll stage	因子分析FN	Y=-271.4FA₁-139.2FA₂+27.89FA₃+3 228	0.735 8	183.5	5.68
	吐絮期 Boll-opening stage	逐步回归SWR	Y=18 792I+1.952D+2 260	0.606 5	224.0	6.94
	吐絮期 Boll-opening stage	因子分析FN	Y=224.8FA₁-112.3FA₂+104.3FA₃+3 252	0.592 6	230.6	7.14
30	花铃期 Flowering and boll stage	逐步回归SWR	Y=-3 085W-3 102E+15 016	0.784 9	165.6	5.13
	花铃期 Flowering and boll stage	因子分析FN	Y=235.2FA₁-134.1FA₂-152.6FA₃+3 228	0.757 4	175.9	5.45
	吐絮期 Boll-opening stage	逐步回归SWR	Y=-296 8W+1 030	0.415 9	272.9	8.45
	吐絮期 Boll-opening stage	FN	Y=204.5FA₁-110FA₂+100.5FA₃	0.502 0	252.0	7.81
40	花铃期 Flowering and boll stage	逐步回归SWR	Y=-238 87M+11 146	0.435 5	268.3	8.31
	花铃期 Flowering and boll stage	因子分析FN	Y=-180.7FA₁-39.06FA₂+158.8FA₃+3 228	0.465 6	261.0	8.09
	吐絮期 Boll-opening stage	逐步回归SWR	Y=-2 369C+19 032	0.506 0	251.0	7.78
	吐絮期 Boll-opening stage	因子分析FN	Y=160.7FA₁+63.53FA₂+165.2FA₃	0.448 1	265.3	8.22

[1]	Chuang LU, Haitang HU, Yuan QIN, Heju HUAI, Cunjun LI. Delineating Management Zones in Spring Maize Field Based on UAV Multispectral Image [J]. Journal of Agricultural Science and Technology, 2022, 24(9): 106-115.
[2]	Lingwen HU, Zhongfa ZHOU, Linjiang YIN, Meng ZHU, Denghong HUANG. Rape Identification at Seedling Stage Based on UAV RGB Image [J]. Journal of Agricultural Science and Technology, 2022, 24(9): 116-128.
[3]	Yangyang LIU, Yue PAN, Shiwei WANG, Haifang HU. Light Response Model Fitting and Comprehensive Evaluation for Vitisamurensis [J]. Journal of Agricultural Science and Technology, 2022, 24(2): 104-114.
[4]	ZHANG Guanhong, WANG Xinjun, XU Xiaolong, YAN Linan, CHANG Mengdi, LI Yongkang, . Desert Vegetation Classification Based on Object-oriented UAV Remote Sensing Images [J]. Journal of Agricultural Science and Technology, 2021, 23(5): 69-77.
[5]	FU Rongtao, CHEN Cheng, WANG Jian, CHEN Xuejuan, LU Daihua*. Control Conditions and Effects of Plant Protection Unmanned Aerial Vehicle (UAV) on Diseases and Insect Pests of Rice#br# [J]. Journal of Agricultural Science and Technology, 2021, 23(4): 103-109.
[6]	ZHOU Kai1, ZHOU Tong1, DING Feng2, DING Dawei2, WU Wei1, YAO Zhaosheng1, LIU Tao1, HUO Zhongyang1, SUN Chengming1*. Wheat LAI Estimation in Main Growth Period Based on UAV Images [J]. Journal of Agricultural Science and Technology, 2021, 23(1): 89-97.
[7]	WANG Bingjie1, PAN Bo1, JIANG Lei1, LIN Yong1*, ZHAO Shuai2, MO Yuxing2. Effects of Spraying Parameters of Plant Protection Unmanned Aerial Vehicle on Deposition Distribution of Droplets in Pitaya Canopy [J]. Journal of Agricultural Science and Technology, 2020, 22(10): 101-109.
[8]	XU Peng1, XU Weicheng2, LUO Yangfan2, ZHAO Zuoxi2*. Precise Classification of Cultivated Land Based on Visible Remote Sensing Image of UAV [J]. Journal of Agricultural Science and Technology, 2019, 21(6): 79-86.
[9]	XU Weicheng1, CHEN Kai1, ZHANG Ming1, LI Zhengfeng2, ZHANG Tianshun1*. Life Lifting Scheme for Plant Protection UAV Based on Solar [J]. Journal of Agricultural Science and Technology, 2018, 20(11): 62-68.
[10]	ZHAO Rongrong1, SHAO Huifang1*, FAN Lei2, ZHANG Manman3, WANG Shanshan1, CHEN Fangquan1, LIU Zhihong1, NIU Guiyan1, HAO Haohao4, XU Zicheng1. Relationship Analasis between Chemical Compositions and Sensory Quality of in Drying Process [J]. Journal of Agricultural Science and Technology, 2017, 19(2): 93-102.
[11]	LI Yan1, CHU Zhiguo2, TIAN Haiying3, HAO Hui3, SONG Jinyong3, ZHOU Hao3, LI Huaiqi3, LIU Pengfei1, ZHAO Mingqin1*. Influences of Organic Acids on Deliveries of Harmful Components in Mainstream Smoke from Flue-curd Tobacco Leaf [J]. Journal of Agricultural Science and Technology, 2017, 19(1): 125-130.
[12]	LIANG Fei, DOU Xue-cheng*. Coupling Degree Analysis of Circular Agriculture in Gansu Province under the Perspective of Ecological Industry Chain and Value Chain [J]. Journal of Agricultural Science and Technology, 2016, 18(2): 200-207.
[13]	AI Hong-juan1,2, JIANG He-ping1*. Studies on Evaluating Modern Agricultural Development Level in Xinjiang based on Factor Analysis Method [J]. , 2015, 17(4): 157-164.
[14]	HUANG Yong-bao1, JI Yan-guang2*. Studies and Analysis on Agricultural Machinery Development in China Based on Factor Analysis [J]. , 2014, 16(4): 95-101.
[15]	NIU Li\\|li1, ZHANG Bao\\|quan2, XU Zi\\|cheng1*, GENG Zong\\|ze3, HUANG Guang\\|hua2. Difference Analysis of Neutral Aroma Component Contents in Flue\\|cured Tobacco Groups with Different Taste Characteristic [J]. , 2014, 16(2): 100-108.