基于RFE_RF算法的幼龄沉香叶片含水率预估模型

doi:10.12302/j.issn.1000-2006.202010043

国家林草科技领军期刊
中国精品科技期刊
中国高校百佳科技期刊
江苏省新闻出版政府奖期刊奖
RCCSE林学权威期刊（A+）
CSCD核心期刊
Scopus数据库收录期刊
中文核心期刊
SCD核心期刊

作者加群：102861116

微信公众号：南京林业大学学报

高级检索

南京林业大学学报（自然科学版） ›› 2022, Vol. 46 ›› Issue (4): 177-184.doi: 10.12302/j.issn.1000-2006.202010043

基于RFE_RF算法的幼龄沉香叶片含水率预估模型

王甜¹(), 王雪峰¹^,^*(), 刘嘉政²

1.中国林业科学研究院资源信息研究所,北京 100091
2.国家林业和草原局森林经营与生长模拟重点实验室,北京 100091

收稿日期:2020-10-26 修回日期:2021-03-09 出版日期:2022-07-30 发布日期:2022-08-01
通讯作者: 王雪峰
基金资助:
国家自然科学基金项目(32071761)

The prediction model of moisture content of young Aquilaria sinensis leaves based on RFE_RF algorithm

WANG Tian¹(), WANG Xuefeng¹^,^*(), LIU Jiazheng²

1. Research Institute of Forest Resources Information Technique, Chinese Academy of Forestry, Beijing 100091
2. Key Laboratory of Forest Management and Growth Modelling, National Forestry and Grassload Administration, Beijing 100091

Received:2020-10-26 Revised:2021-03-09 Online:2022-07-30 Published:2022-08-01
Contact: WANG Xuefeng

摘要/Abstract

摘要：

【目的】针对随机森林算法在树木水分预测模型中高维度变量筛选困难及精度较低的问题,研究基于递归特征消除(RFE)与随机森林(RF)的融合算法,构建幼龄沉香(Aquilaria sinensis)可见光图像与叶片含水率的估测模型,探索适合幼龄沉香生长的水分条件,为实现沉香幼苗水分亏缺程度的无损监测提供可行方法。【方法】以2年生的名贵树种沉香为研究对象,用相机获取4种不同水分梯度下的幼龄沉香可见光图像,提取15种图像特征,利用递归特征消除法筛选沉香叶片最优的图像特征子集,然后结合随机森林算法构建沉香叶片含水率的预测模型,最后利用十折交叉验证法,将RFE_RF模型与常规随机森林(RF)以及最小二乘法支持向量机(LSSVM)相比较,检验模型的可行性。利用递归特征消除和随机森林融合(RFE_RF)算法筛选出幼龄沉香叶片图像的标准红光值(I_NR)、饱和度(S)、矩形度(E_R)3个特征,并以此作为模型自变量。【结果】与重度水淹胁迫相比,幼龄沉香对于长期重度干旱胁迫更加敏感,且干旱时间超出2周时幼苗叶片严重受损,威胁沉香生长;沉香最适叶片水分生长范围为50%~65%,适度增加水分,有利于沉香生长。基于RFE_RF融合算法构建的预测模型敏感度、特异性、误报率和精度分别达到88.64%、85. 31%、14. 39%和91.62%,优于LSSVM模型效果;与RF预测模型相比其敏感度提高3.34%、特异性提高10.87%、误报率降低36.83%、精度提高13.39%。【结论】基于RFE_RF融合算法建立的沉香叶片颜色、形状特征与含水率的模型,解决了随机森林过程中高维度变量选择问题,提高了RF在林木水分预测模型中的精度,实现了沉香幼苗叶片含水率的无损估测和诊断,为珍贵树种在经营管理中对水分进行准确管控提供了新思路。

关键词: 沉香, 递归特征消除(RFE), 随机森林(RF)算法, 含水率预估

Abstract:

【Objective】 To solve the problems of difficult selections of high-dimensional variables and low accuracy of random forest algorithm in tree water prediction model, this paper studies the fusion algorithm based on recursive feature elimination (RFE) and random forest (RF), constructs the estimation model of visible light images and leaf water content of young A. sinensis, explores the water conditions suitable for the growth of young Aquilaria sinensis, and provides a feasible reference method for realizing the nondestructive monitoring of water deficit degree of A. sinensis seedlings. 【Method】 Taking two-years-old precious tree species of A. sinensis as the research object, the visible light images of young A. sinensis under four different water gradients were obtained by camera, 15 image features were extracted, the recursive feature elimination method was used to screen the optimal image feature subset of agallochum leaves, and then the prediction model of water content of A. sinensis leaves was constructed combined with RF algorithm. Finally, the RFE_RF model is compared with the conventional RF and least square support vector machine (LSSVM) to test the feasibility of the model. Three features of standard red light value (I_NR), saturation (S) and rectangularity (E_R) of young A. sinensis leaf images were screened out by recursive feature elimination and random forest fusion (RFE_RF) algorithm, and used as the model independent variables. 【Result】 The results showed that compared with severe flooding stress, young A. sinensis was more sensitive to long-term severe drought stress, and the drought time exceeded two weeks, the leaves of the seedlings were severely damaged, threatening the growth of A. sinensis; the optimum leaf water growth range of A. sinensis was 50%-65%, moderately increase moisture, which is conducive to the growth of A. sinensis. The false alarm rate and accuracy of the prediction model constructed based on the RFE_RF fusion algorithm reached 88.64%, 85.31%, 14.39% and 91.62%, respectively, which were better than the LSSVM model; the sensitivity is increased by 3.34%, the specificity is increased by 10.87%, the false alarm rate is decreased by 36.83%, and the precision is increased by 13.39%. 【Conclusion】 The model of Aquilaria sinensis leaf color, shape characteristics and water content established by RFE_RF solves the problem of high-dimensional variable selections in the random forest process, improves the accuracy of RF in the forest moisture prediction model, and realizes the analysis of leaf water content of Aquilaria sinensis seedlings. The non-destructive estimation and diagnosis provide new ideas for accurate water management and management of precious tree species.

Key words: Aquilaria sinensis, recursive feature elimination(RFE), random forest (RF) algorithm, moisture content estimation

中图分类号:

TP181
S718

王甜,王雪峰,刘嘉政. 基于RFE_RF算法的幼龄沉香叶片含水率预估模型[J]. 南京林业大学学报（自然科学版）, 2022, 46(4): 177-184.

WANG Tian, WANG Xuefeng, LIU Jiazheng. The prediction model of moisture content of young Aquilaria sinensis leaves based on RFE_RF algorithm[J].Journal of Nanjing Forestry University (Natural Science Edition), 2022, 46(4): 177-184.DOI: 10.12302/j.issn.1000-2006.202010043.

图/表 9

表1

图1

图2

图3

图4

图5

图6

图7

表2

参考文献 29

[1]	宋晓琛, 王西洋, 杨光, 等. 无机盐与激素混合对土沉香结香的诱导[J]. 林业科学, 2020, 56(8):121-130.
	SONG X C, WANG X Y, YANG G, et al. Mechanism of agarwood formation under the induction of both inorganic salts and hormones[J]. Sci Silvae Sin, 2020, 56(8):121-130.DOI:10.11707/j.1001-7488.20200814.
[2]	张沛健, 尚秀华, 吴志华. 基于图像处理技术的5种红树林叶片形态特征及叶绿素相对含量的估测[J]. 热带作物学报, 2020, 41(3):496-503.
	ZHANG P J, SHANG X H, WU Z H. Morphological characteristics of leaves and estimation of relative chlorophyll content in five species of mangrove based on image processing technology[J]. Chin J Trop Crops, 2020, 41(3):496-503.DOI: 10.3969/j.issn.1000-2561.2020.03.011.
[3]	MIGLIACCIO K W, SCHAFFER B, CRANE J H, et al. Plant response to evapotranspiration and soil water sensor irrigation scheduling methods for Papaya production in south Florida[J]. Agric Water Manag, 2010, 97(10):1452-1460.DOI:10.1016/j.agwat.2010.04.012.
[4]	KACIRA M, LING P P, SHORT T H. Machine vision extracted plant movement for early detection of plant water stress[J]. Trans ASAE, 2002, 45(4): 1147-1153.DOI:10.13031/2013.9923.
[5]	AHMAD U, MARDISON S, TJAHJOHUTOMO R, et al. Development of automatic grading machine prototype for Citrus using image processing[J]. Aust J Agric Eng, 2010, 1(5):165-169.
[6]	ZAKALUK R, SRI RANJAN R. Artificial neural network modelling of leaf water potential for potatoes using RGB digital images: a greenhouse study[J]. Potato Res, 2006, 49(4):255-272.DOI:10.1007/s11540-007-9021-x.
[7]	劳东青, 李发永, 曹洪武. 图像去噪及分割算法在枣叶含水率估算中的影响研究[J]. 节水灌溉, 2018(6):1-6.
	LAO D Q, LI F Y, CAO H W. Influence of image denoising and image segmentation on estimating leaf water content of jujube[J]. Water Sav Irrigation, 2018(6):1-6.DOI: 10.3969/j.issn.1007-4929.2018.06.002.
[8]	张娟利, 宋朝阳, 韩文霆, 等. 基于RGB图像处理的烟叶水分无损检测方法研究[J]. 中国农机化学报, 2019, 40(5):62-68.
	ZHANG J L, SONG Z Y, HAN W T, et al. Nondestructive testing of tobacco leaf water status by using digital RGB images[J]. J Chin Agric Mech, 2019, 40(5)62-68.DOI:10.13733/j.jcam.issn.2095-5553.2019.05.12.
[9]	LI L, USTIN S L, RIANO D. Retrieval of fresh leaf fuel moisture content using genetic algorithm partial least squares (GA-PLS) modeling[J]. IEEE Geosci Remote Sens Lett, 2007, 4(2):216-220.DOI:10.1109/LGRS.2006.888847.
[10]	江朝晖, 杨春合, 周琼, 等. 基于图像特征的越冬期冬小麦冠层含水率检测[J]. 农业机械学报, 2015, 46(12):260-267.
	JIANG Z H, YANG C H, ZHOU Q. et al. Detection of canopy water content of winter wheat during wintering period based on image features[J]. Trans Chin Soc Agric Mach, 2015, 46(12): 260-267.DOI: 10.6041/j.issn.1000-1298.2015.12.035.
[11]	VESALI F, GHARIBKHANI M, KOMARIZADEH M H. An approach to estimate moisture content of apple with image processing method[J]. Aust J Crop Sci, 2011, 5:111-115.
[12]	顾金梅, 吴雪梅, 龙曾宇, 等. 基于BP神经网络的烟叶颜色自动分级研究[J]. 中国农机化学报, 2016(4):110-114.
	GU J M, WU X M, LONG Z Y, et al. Automatic classification research on tobacco leaf color based on BP neural network[J]. J Chin Agric Mech, 2016(4):110-114.
[13]	陈珠琳, 王雪峰. 水胁迫下多角度幼龄檀香图像颜色变化分析及含水率反演[J]. 应用生态学报, 2019, 30(8):2639-2646.
	CHEN Z L, WANG X F. Color change analysis and water content inversion of young sandalwood in multi-angle under water stress[J]. Chin J Appl Ecol, 2019, 30(8)2639-2646.DOI:10.13287/j.1001-9332.201908.024.
[14]	张海威, 张飞, 张贤龙, 等. 光谱指数的植被叶片含水量反演[J]. 光谱学与光谱分析, 2018, 38(5):1540-1546.
	ZHANG H W, ZHANG F, ZHANG X L, et al. Inversion of vegetation leaf water content based on spectral index[J]. Spectrosc Spectr Anal, 2018, 38(5):1540-1546.
[15]	宋镇, 姬长英, 张波. 基于高光谱技术融合图像信息的杏鲍菇干燥过程中含水率检测[J]. 江苏农业学报, 2019, 35(2):436-444.
	SONG Z, JI C Y, ZHANG B. Visualized determination of moisture content in Pleurotus eryngii during drying process based on hyperspectral imaging technology[J]. Jiangsu J Agric Sci, 2019, 35(2):436-444.DOI:10.3969/j.issn.1000-4440.2019.02.026.
[16]	陈珠琳, 王雪峰, 孙汉中. 基于可见光-近红外图像的幼龄檀香全磷含量诊断[J]. 北京林业大学学报, 2019, 41(2):88-96.
	CHEN Z L, WANG X F, SUN H Z. Diagnosis of total phosphorus content in young sandalwood based on visible light and near infrared images[J]. J Beijing For Univ, 2019, 41(2):88-96.DOI:10.13332/j.1000-1522.20180214.
[17]	孙媛媛. 基于水稻叶片图像时空动态特征的氮磷钾营养诊断[D]. 杭州: 浙江大学, 2018.
	SUN Y Y. Diagnosis of nitrogen phosphorus and potassium deficiency based on temporal dynamics of rice leaf image[D]. Hangzhou: Zhejiang University, 2018.
[18]	吴辰文, 梁靖涵, 王伟, 等. 基于递归特征消除方法的随机森林算法[J]. 统计与决策, 2017(21):60-63.
	WU C W, LIANG J H, WANG W, et al. Random forest algorithm based on recursive feature elimination[J]. Stat Decis, 2017(21):60-63.DOI:10.13546/j.cnki.tjyjc.2017.21.014.
[19]	刘笑笑. 基于RF-RFE算法的森林生物量遥感特征选择方法研究[D]. 泰安: 山东农业大学, 2016.
	LIU X X. Study on the remote sensing feature selection method for forest biomass estimation based on RF-RFE[D]. Taian: Shandong Agricultural University, 2016.
[20]	魏小敏, 徐彬, 关佶红. 基于递归特征消除法的蛋白质能量热点预测[J]. 山东大学学报(工学版), 2014, 44(2):12-20.
	WEI X M, XU B, GUAN J H. Prediction of protein energy hot spots based on recursion feature elimination[J]. J Shandong Univ (Eng Sci), 2014, 44(2):12-20.DOI: 10. 6040/j.issn.1672-3961.1.2013.378.
[21]	BREIMAN L. Random forests[J]. Machine Learning, 2001, 45(1):5-32.
[22]	杜学惠, 孟春, 刘美爽. 基于单个特征分类准确率的特征选择方法研究[J]. 南京林业大学学报(自然科学版), 2019, 43(4):109-116.
	DU X H, MENG C, LIU M S. Research on feature selection based on single feature classification accuracy[J]. J Nanjing For Univ (Nat Sci Ed), 2019, 43(4):109-116.DOI: 10.3969/j.issn.1000-2006.201807059.
[23]	SUYKENS J A K, VANDEWALLE J. Least squares support vector machine classifiers[J]. Neural Process Lett, 1999, 9(3):293-300.DOI:10.1023/A:1018628609742.
[24]	田宝. 干旱胁迫对大叶黄杨保护酶及渗透物质含量的影响[J]. 山东林业科技, 2019, 49(4):64-67.
	TIAN B. Effects of drought stress on protective enzymes and osmotic substances of Buxus megistophylla levl.[J]. J Shandong For Sci Technol, 2019, 49(4):64-67.DOI: 10.3969/j.issn.1002-2724.2019.04.017.
[25]	刘亮, 郝立华, 李菲, 等. CO₂浓度和温度对玉米光合性能及水分利用效率的影响[J]. 农业工程学报, 2020, 36(5):122-129.
	LIU L, HAO L H, LI F, et al. Effects of CO₂ concentration and temperature on leaf photosynthesis and water use efficiency in maize[J]. Trans Chin Soc Agric Eng, 2020, 36(5): 122-129.DOI: 10.11975/j.issn.1002-6819.2020.05.014.
[26]	张森茂. 灌溉和施氮对楸树、云杉幼树生长和光合的影响[D]. 杨凌: 西北农林科技大学, 2019.
	ZHANG S M. Effects of irrigation and nitrogen application on growth and photosynthesis of Catalpa bungei and Picea saplings[D]. Yangling: Northwest A & F University, 2019.
[27]	吴顺, 张雪芹, 蔡燕. 干旱胁迫对黄瓜幼苗叶绿素含量和光合特性的影响[J]. 中国农学通报, 2014, 30(1):133-137.
	WU S, ZHANG X Q, CAI Y. Effects of drought stress on chlorophyll contents and photosynthetic characteristics of cucumber seedlings[J]. Chin Agric Sci Bull, 2014, 30(1):133-137.
[28]	张世柯, 黄耀, 简曙光, 等. 热带滨海植物红厚壳的抗逆生物学特性[J]. 热带亚热带植物学报, 2019, 27(4):391-398.
	ZHANG S K, HUANG Y, JIAN S G, et al. Stress resistance characteristics of Calophyllum inophyllum,a tropical beach plant[J]. J Trop Subtrop Bot, 2019, 27(4):391-398.DOI:10.11926/jtsb.4043.
[29]	李婕, 刘楠, 任海, 等. 7种植物对热带珊瑚岛环境的生态适应性[J]. 生态环境学报, 2016, 25(5):790-794.
	LI J, LIU N, REN H, et al. Ecological adaptability of seven plant species to tropical coral island habitat[J]. Ecol Environ Sci, 2016, 25(5):790-794.DOI: 10.16258/j.cnki.1674-5906.2016.05.009.

处理 treatment	含水率/% water content
处理 treatment	1	2	3	4	5	6	7	8	9	10	11
CK	58.8	59.2	59.6	60.1	60.8	61.9	64.1	64.3	65.0	65.3	65.4
T3	45.9	59.5	60.2	60.4	62.3	64.2	66.5	64.6	65.1	66.1	65.9
T2	60.3	61.4	62.4	63.5	65.5	67.3	68.5	69.3	70.6	71.0	65.7
T1	64.1	63.4	64.8	68.9	65.6	66.2	66.3	67.4	75.1	69.5	70.5

算法 algorithm	敏感度 sensitivity	特异性 specificity	误报率 false positive rate	精度 precision of measurement	排序 rank
LSSVM	0.843 2	0.750 0	0.234 5	0.742 1	3
RF	0.857 8	0.769 4	0.227 8	0.764 0	2
RFE_RF	0.886 5	0.853 1	0.143 9	0.916 2	1

基于RFE_RF算法的幼龄沉香叶片含水率预估模型

The prediction model of moisture content of young Aquilaria sinensis leaves based on RFE_RF algorithm

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 9

参考文献 29

相关文章 2

编辑推荐

Metrics

本文评价

作者

读者

审稿

[1]	袁莹, 王雪峰, 王甜, 陈飞飞, 黄川腾, 林玲, 董晓娜. 基于多图像特征的幼龄沉香全氮估测[J]. 南京林业大学学报（自然科学版）, 2023, 47(3): 19-28.
[2]	高家军, 张旭, 郭颖, 刘昱坤, 郭安琪, 石蒙蒙, 王鹏, 袁莹. 融合Swin Transformer的虫害图像实例分割优化方法研究[J]. 南京林业大学学报（自然科学版）, 2023, 47(3): 1-10.