基于高光谱和深度迁移学习的柑橘叶片钾含量反演
|
摘要
针对传统柑橘叶片钾含量检测方法耗时费力、操作繁琐且损伤叶片等弊端,引入高光谱信息探索柑橘叶片钾含量快速无损检测与预测模型,选用ASD Field Spec 3光谱仪采集柑橘4个重要物候期(萌芽期、稳果期、壮果促梢期和采果期)的叶片反射光谱,同步采用火焰光度法测定叶片的钾含量;先用正交试验确定小波去噪的最佳去噪参数组合,再进行不同光谱形式变换,对不同物候期光谱进行基于堆栈稀疏编码机-深度学习网络(Stacked sparse autoencoder-deep learning networks,SSAE-DLNs)的特征提取迁移和融合多种特征,对比支持向量机回归、偏最小二乘法回归、广义神经网络、逐步多元线性回归等多种诊断模型,结果表明,模型SSAE-DLNs基于一阶微分光谱特征建立全生长期钾含量预测模型的性能最优,其校正集和验证集决定系数分别为0. 898 8、0. 877 1,均方根误差分别为0. 544 3、0. 552 8。试验表明,深度迁移学习网络可对柑橘叶片钾含量进行精确预测,为高光谱检测技术用于柑橘树长势监测和营养诊断提供了参考。
Traditional methods of obtaining potassium content of citrus leaves are time-consuming procedures with complex operations which can be harmful to citrus trees. Moreover,traditional methods cannot meet the demand for rapid and non-destructive monitoring of potassium content in large-scale citrus orchards. Combined with the state-of-the-art deep learning technology,a model based on stacked sparse autoencoder( SSAE) and deep learning networks( DLNs) using hyperspectral information for potassium content prediction in four growth stages was proposed. The experiments were conducted in the Crab Village of Luogang District,Guangzhou City,Guangdong Province,and the samples were 109 citrus trees planted. During four growth stages, i. e., germination, stability, bloom and picking stages,hyperspectral reflectance of citrus leaves was respectively measured by spectrometer( ASD FieldSpec 3),and at the same time,potassium content of citrus leaves was obtained by using traditional chemical method. All the collected samples constituted a large-scale dataset with totally 436 tuples,80% of which were utilized as the calibration set and remaining 20% as the validation set. The constructed model which relied on the calibration set and the validation set was evaluated respectively. Firstly, successive projection algorithm( SPA) was provided to deal with the high-dimensional spectral vectors for dimension reduction and feature extraction. A prediction model of multiple linear regression( MLR) for potassium content of citrus leaves was established based on those extracted features. The result showed that the potassium spectrum contained a large number of complex nonlinear characteristics. Secondly,wavelet denoising was applied to reduce the high-frequency noise in the original spectrum,and the optimized parameter combination of wavelet de-noising through orthogonal test was as follows: "coif2 "as wavelet basis function,the number of decomposition layer was 3,"sqtwolog"as the threshold,and "one"as noise estimation scheme,respectively. Thirdly,the features of SSAE in a specific stage were transferred and merged into baseline layer by layer to find out the best number of layers. The result showed that the best numbers of transferred layer were 3,1,4 and 3,and the corresponding values of determination coefficients for calibration set were 0. 899 9,0. 859 8,0. 886 9 and 0. 854 7 at germination,stability,bloom and picking stages,respectively,which were improved by 19. 82%,9. 45%,21. 49% and7. 21%,respectively,compared with baseline. Then,the features of SSAE in the best layer were transferred and merged into baseline stage by stage to find out the best number of transferred stage. The experiment revealed that features of all four stages were transferred to its corresponding stage domain achieving the best performance. In this situation,the coefficients of determination for calibration set were0. 877 2,0. 898 1,0. 904 9 and 0. 889 4 at germination, stability, bloom and picking stages,respectively,which were improved by 16. 80%, 14. 32%, 23. 96% and 11. 56%, respectively,compared with baseline. Fourthly,after performing wavelet de-noising and four kinds of spectrum transformation,i. e.,the first derivative,second derivative,reciprocal and logarithm to the original spectrum,the layers' features and stages' features,which were obtained in SSAE previously,were transferred and merged into spectrum in four growth stages. When the first derivative spectrum was used as the input vector of the samples with wavelet de-noising,the SSAE-DLNs model achieved the best result and the coefficients of determination for calibration set were 0. 899 2,0. 889 9,0. 883 8,0. 872 7 and0. 898 8,respectively,and the corresponding values of RMSE were 0. 542 5,0. 549 6,0. 550 9,0. 5539 and 0. 544 3,respectively,and the corresponding values of sparse proportion were 0. 141 1,0. 163 3,0. 118 9,0. 185 6 and 0. 207 8,respectively,at germination,stability,bloom,picking stages and the whole growth period; and for the validation set,the coefficients of determination were 0. 865 1,0. 870 4,0. 855 1,0. 858 0 and 0. 877 1,respectively,and the corresponding values of RMSE were 0. 569 3,0. 567 4,0. 578 6,0. 572 2 and 0. 552 8,respectively. Comparing with traditional models such as support vector regression( SVR),partial least square regression( PLSR),general regression neural networks( GRNN) and stepwise multiple linear regression( SMLR),SSAE-DLNs model achieved the best performance,and the next was SVR,in which R~2 of calibration and validation set were 0. 898 8 and0. 877 1,respectively. Finally,the research result proved the feasibility of monitoring potassium content of citrus leaves,which may provide a theoretical basis for growth monitoring and nutritional diagnosis of citrus trees.
Traditional methods of obtaining potassium content of citrus leaves are time-consuming procedures with complex operations which can be harmful to citrus trees. Moreover,traditional methods cannot meet the demand for rapid and non-destructive monitoring of potassium content in large-scale citrus orchards. Combined with the state-of-the-art deep learning technology,a model based on stacked sparse autoencoder( SSAE) and deep learning networks( DLNs) using hyperspectral information for potassium content prediction in four growth stages was proposed. The experiments were conducted in the Crab Village of Luogang District,Guangzhou City,Guangdong Province,and the samples were 109 citrus trees planted. During four growth stages, i. e., germination, stability, bloom and picking stages,hyperspectral reflectance of citrus leaves was respectively measured by spectrometer( ASD FieldSpec 3),and at the same time,potassium content of citrus leaves was obtained by using traditional chemical method. All the collected samples constituted a large-scale dataset with totally 436 tuples,80% of which were utilized as the calibration set and remaining 20% as the validation set. The constructed model which relied on the calibration set and the validation set was evaluated respectively. Firstly, successive projection algorithm( SPA) was provided to deal with the high-dimensional spectral vectors for dimension reduction and feature extraction. A prediction model of multiple linear regression( MLR) for potassium content of citrus leaves was established based on those extracted features. The result showed that the potassium spectrum contained a large number of complex nonlinear characteristics. Secondly,wavelet denoising was applied to reduce the high-frequency noise in the original spectrum,and the optimized parameter combination of wavelet de-noising through orthogonal test was as follows: "coif2 "as wavelet basis function,the number of decomposition layer was 3,"sqtwolog"as the threshold,and "one"as noise estimation scheme,respectively. Thirdly,the features of SSAE in a specific stage were transferred and merged into baseline layer by layer to find out the best number of layers. The result showed that the best numbers of transferred layer were 3,1,4 and 3,and the corresponding values of determination coefficients for calibration set were 0. 899 9,0. 859 8,0. 886 9 and 0. 854 7 at germination,stability,bloom and picking stages,respectively,which were improved by 19. 82%,9. 45%,21. 49% and7. 21%,respectively,compared with baseline. Then,the features of SSAE in the best layer were transferred and merged into baseline stage by stage to find out the best number of transferred stage. The experiment revealed that features of all four stages were transferred to its corresponding stage domain achieving the best performance. In this situation,the coefficients of determination for calibration set were0. 877 2,0. 898 1,0. 904 9 and 0. 889 4 at germination, stability, bloom and picking stages,respectively,which were improved by 16. 80%, 14. 32%, 23. 96% and 11. 56%, respectively,compared with baseline. Fourthly,after performing wavelet de-noising and four kinds of spectrum transformation,i. e.,the first derivative,second derivative,reciprocal and logarithm to the original spectrum,the layers' features and stages' features,which were obtained in SSAE previously,were transferred and merged into spectrum in four growth stages. When the first derivative spectrum was used as the input vector of the samples with wavelet de-noising,the SSAE-DLNs model achieved the best result and the coefficients of determination for calibration set were 0. 899 2,0. 889 9,0. 883 8,0. 872 7 and0. 898 8,respectively,and the corresponding values of RMSE were 0. 542 5,0. 549 6,0. 550 9,0. 5539 and 0. 544 3,respectively,and the corresponding values of sparse proportion were 0. 141 1,0. 163 3,0. 118 9,0. 185 6 and 0. 207 8,respectively,at germination,stability,bloom,picking stages and the whole growth period; and for the validation set,the coefficients of determination were 0. 865 1,0. 870 4,0. 855 1,0. 858 0 and 0. 877 1,respectively,and the corresponding values of RMSE were 0. 569 3,0. 567 4,0. 578 6,0. 572 2 and 0. 552 8,respectively. Comparing with traditional models such as support vector regression( SVR),partial least square regression( PLSR),general regression neural networks( GRNN) and stepwise multiple linear regression( SMLR),SSAE-DLNs model achieved the best performance,and the next was SVR,in which R~2 of calibration and validation set were 0. 898 8 and0. 877 1,respectively. Finally,the research result proved the feasibility of monitoring potassium content of citrus leaves,which may provide a theoretical basis for growth monitoring and nutritional diagnosis of citrus trees.
引文
[1]黄双萍,岳学军,洪添胜,等.不同物候期柑橘叶片钾素水平预测建模[J].江苏大学学报(自然科学版),2013,34(5):529-535.HUANG Shuangping,YUE Xuejun,HONG Tiansheng,et al. Potassium content prediction model of citrus leaves in different phenological period[J]. Journal of Jiangsu University(Natural Science Edition),2013,34(5):529-535.(in Chinese)
[2]章胜勇,祁春节.我国柑橘鲜果滞销的原因及对策分析[J].华南农业大学学报(社会科学版),2008,7(4):75-83.ZHANG Shengyong,QI Chunjie. Analysis of the causes and countermeasures against the slow sale of citrus fruit in China[J].Journal of South China Agricultural University(Social Science Edition),2008,7(4):75-83.(in Chinese)
[3] LIU Fei,NIE Pengcheng,HUANG Min,et al. Nondestructive determination of nutritional information in oilseed rape leaves using visible/near infrared spectroscopy and multivariate calibrations[J]. Science China Information Sciences,2011,54(3):598-608.
[4] ABDEL-RAHMAN E M,AHMED F B,ISMAIL R. Random forest regression and spectral band selection for estimating sugarcane leaf nitrogen concentration using EO-1 Hyperion hyperspectral data[J]. International Journal of Remote Sensing,2013,34(2):712-728.
[5]黄双萍,洪添胜,岳学军,等.基于高光谱的柑橘叶片氮素含量多元回归分析[J].农业工程学报,2013,29(5):132-138.HUANG Shuangping,HONG Tiansheng,YUE Xuejun,et al. Multiple regression analysis of citrus leaf nitrogen content using hyperspectral technology[J]. Transactions of the CSAE,2013,29(5):132-138.(in Chinese)
[6] TONG Peijin,DU Yiping,ZHENG Kaiyi,et al. Improvement of NIR model by fractional order Savitzky-Golay derivation(FOSGD)coupled with wavelength selection[J]. Chemometrics and Intelligent Laboratory Systems,2015,143:40-48.
[7] MENESATTI P,ANTONUCCI F,PALLOTTINO F,et al. Estimation of plant nutritional status by vis-NIR spectrophotometric analysis on orange leaves[J]. Biosystems Engineering,2010,105(4):448-454.
[8]刘艳丽,何绍兰,吕强,等.柑橘花钾素营养的高光谱表征[J].果树学报,2014,31(6):1065-1071.LIU Yanli,HE Shaolan,LQiang,et al. Characteristics of potassium content in citrus flowers with hyperspectral imagery[J].Journal of Fruit Science,2014,31(6):1065-1071.(in Chinese)
[9]朱西存,姜远茂,赵庚星,等.基于模糊识别的苹果花期冠层钾素含量高光谱估测[J].光谱学与光谱分析,2013,33(4):1023-1027.ZHU Xicun,JIANG Yuanmao,ZHAO Gengxing,et al. Hyperspectral estimation of kalium content in apple florescence canopy based on fuzzy recognition[J]. Spectroscopy and Spectral Analysis,2013,33(4):1023-1027.(in Chinese)
[10]赵化兵,王洁,董彩霞,等.基于可见/近红外反射光谱的梨树叶片钾含量的快速测定研究[J].土壤,2014,46(2):256-261.ZHAO Huabing,WANG Jie,DONG Caixia,et al. Fast determination of potassium content in fresh pear leaves using visible/near infrared spectroscopy technique[J]. Soils,2014,46(2):256-261.(in Chinese)
[11]屈莎,李振海,邱春霞,等.基于开花期氮素营养指标的冬小麦籽粒蛋白质含量遥感预测[J].农业工程学报,2017,33(12):186-193.QU Sha,LI Zhenhai,QIU Chunxia,et al. Remote sensing prediction of winter wheat grain protein content based on nitrogen nutrition index at anthesis stage[J]. Transactions of the CSAE,2017,33(12):186-193.(in Chinese)
[12] PAN S J,YANG Qiang. A survey on transfer learning[J]. IEEE Transactions on Knowledge and Data Engineering,2010,22(10):1345-1359.
[13] BARNETT S M,CECI S J. When and where do we apply what we learn? A taxonomy for far transfer[J]. Psychological Bulletin,2002,128(4):612-637.
[14] LU Jie,BEHBOOD V,HAO Peng,et al. Transfer learning using computational intelligence:a survey[J]. Knowledge-Based Systems,2015,80(Supp.):14-23.
[15]岳学军,全东平,洪添胜,等.不同生长期柑橘叶片磷含量的高光谱预测模型[J].农业工程学报,2015,31(8):207-213.YUE Xuejun,QUAN Dongping,HONG Tiansheng,et al. Prediction model of phosphorus content for citrus leaves during different growth periods based on hyperspectrum[J]. Transactions of the CSAE,2015,31(8):207-213.(in Chinese)
[16]岳学军,全东平,洪添胜,等.柑橘叶片叶绿素含量高光谱无损检测模型[J].农业工程学报,2015,31(1):294-302.YUE Xuejun,QUAN Dongping,HONG Tiansheng,et al. Non-destructive hyperspectral measurement model of chlorophyll content for citrus leaves[J]. Transactions of the CSAE,2015,31(1):294-302.(in Chinese)
[17] MAHAJAN G R,SAHOO R N,PANDEY R N,et al. Using hyperspectral remote sensing techniques to monitor nitrogen,phosphorus,sulphur and potassium in wheat(Triticum aestivum L.)[J]. Precision Agriculture,2014,15(5):499-522.
[18] GOMEZ-CASERO M T, LOPEZ-GRANADOS F, PENA-BARRAGAn J M, et al. Assessing nitrogen and potassium deficiencies in olive orchards through discriminant analysis of hyperspectral data[J]. Journal of the American Society for Horticultural Science,2007,132(5):611-618.
[19] WU Di,WU Hongxi,CAI Jingbo,et al. Classifying the species of exopalaemon by using visible and near infrared spectra with uninformative variable elimination and successive projections algorithm[J]. Journal of Infrared and Millimeter Waves,2009,28(6):423-427.
[20]邓小蕾,李民赞,郑立华,等.基于反射光谱预处理的苹果叶片叶绿素含量预测[J].农业工程学报,2014,30(14):140-147.DENG Xiaolei,LI Minzan,ZHENG Lihua,et al. Estimating chlorophyll content of apple leaves based on preprocessing of reflectance spectra[J]. Transactions of the CSAE,2014,30(14):140-147.(in Chinese)
[21] TAO Chao,PAN Hongbo,LI Yansheng,et al. Unsupervised spectral-spatial feature learning with stacked sparse autoencoder for hyperspectral imagery classification[J]. IEEE Geoscience and Remote Sensing Letters,2015,12(12):2438-2442.
[22] ZABALZA J,REN J,ZHENG Jiangbin,et al. Novel segmented stacked autoencoder for effective dimensionality reduction and feature extraction in hyperspectral imaging[J]. Neurocomputing,2016,185:1-10.
[23] ZHAO Chunhui,WAN Xiaoqing,ZHAO Genping,et al. Spectral-spatial classification of hyperspectral imagery based on stacked sparse autoencoder and random forest[J]. European Journal of Remote Sensing,2017,50(1):47-63.
[24] HINTON G E,SALAKHUTDINOV R R. Reducing the dimensionality of data with neural networks[J]. Science,2006,313(5786):504-507.
[25] LECUN Y,BENGIO Y,HINTON G. Deep learning[J]. Nature,2015,521(7553):436-444.
[26]岳学军,凌康杰,洪添胜,等.基于高光谱图像的龙眼叶片叶绿素含量分布模型[J/OL].农业机械学报,2018,49(8):18-25.YUE Xuejun,LING Kangjie,HONG Tiansheng,et al. Distribution model of chlorophyll content for longan leaves based on hyperspectral imaging technology[J/OL]. Transactions of the Chinese Society for Agricultural Machinery,2018,49(8):18-25. http:∥www. j-csam. org/jcsam/ch/reader/view_abstract. aspx? flag=1&file_no=20180802&journal_id=jcsam. DOI:10. 6041/j. issn. 1000-1298. 2018. 08. 002.(in Chinese)
[27]孙俊,丛孙丽,毛罕平,等.基于高光谱的油麦菜叶片水分CARS-ABC-SVR预测模型[J].农业工程学报,2017,33(5):178-184.SUN Jun,CONG Sunli,MAO Hanping,et al. CARS-ABC-SVR model for predicting leaf moisture of leaf-used lettuce based on hyperspectral[J]. Transactions of the CSAE,2017,33(5):178-184.(in Chinese)
[28] ZHOU Weixun,SHAO Zhenfeng,DIAO Chunyuan,et al. High-resolution remote-sensing imagery retrieval using sparse features by auto-encoder[J]. Remote Sensing Letters,2015,6(10):775-783.
[29]王珂,沈掌泉,ABOU-ISMAIL O,等.不同钾营养水平的水稻冠层和叶片光谱特征研究初报[J].科技通报,1997,13(4):211-214.WANG Ke,SHEN Zhangquan,ABOU-ISMAIL O,et al. Preliminary study on canopy and leaf reflectance characteristics of rice with various potassium levels[J]. Bulletin of Science and Technology,1997,13(4):211-214.(in Chinese)
[30]刘延,吴秋菊,舒清态.基于非成像高光谱的云烟97号叶片钾含量估算研究[J].山东农业科学,2018,50(4):138-141.LIU Yan,WU Qiuju,SHU Qingtai. Estimation study on potassium content in Yunyan 97 leaves based on non-imaging hyperspectral data[J]. Shandong Agricultural Sciences,2018,50(4):138-141.(in Chinese)
[31] ZHONG Ping,WANG Runsheng. Jointly learning the hybrid CRF and MLR model for simultaneous denoising and classification of hyperspectral imagery[J]. IEEE Transactions on Neural Networks and Learning Systems,2014,25(7):1319-1334.
[32]刘炜,常庆瑞,郭曼,等.土壤导数光谱小波去噪与有机质吸收特征提取[J].光谱学与光谱分析,2011,31(1):100-104.LIU Wei,CHANG Qingrui,GUO Man,et al. Extraction of first derivative spectrum features of soil organic matter via wavelet de-noising[J]. Spectroscopy and Spectral Analysis,2011,31(1):100-104.(in Chinese)
[33]梁亮,杨敏华,臧卓.基于小波去噪与SVR的小麦冠层含氮率高光谱测定[J].农业工程学报,2010,26(12):248-253.LIANG Liang,YANG Minhua,ZANG Zhuo. Determination of wheat canopy nitrogen content ratio by hyperspectral technology based on wavelet denoising and support vector regression[J]. Transactions of the CSAE,2010,26(12):248-253.(in Chinese)
[34]张艳超,肖宇钊,庄载椿,等.基于小波分解的油菜多光谱图像与深度图像数据融合方法[J].农业工程学报,2016,32(16):143-150.ZHANG Yanchao,XIAO Yuzhao,ZHUANG Zaichun,et al. Data fusion of multispectral and depth image for rape plant based on wavelet decomposition[J]. Transactions of the CSAE,2016,32(16):143-150.(in Chinese)
[2]章胜勇,祁春节.我国柑橘鲜果滞销的原因及对策分析[J].华南农业大学学报(社会科学版),2008,7(4):75-83.ZHANG Shengyong,QI Chunjie. Analysis of the causes and countermeasures against the slow sale of citrus fruit in China[J].Journal of South China Agricultural University(Social Science Edition),2008,7(4):75-83.(in Chinese)
[3] LIU Fei,NIE Pengcheng,HUANG Min,et al. Nondestructive determination of nutritional information in oilseed rape leaves using visible/near infrared spectroscopy and multivariate calibrations[J]. Science China Information Sciences,2011,54(3):598-608.
[4] ABDEL-RAHMAN E M,AHMED F B,ISMAIL R. Random forest regression and spectral band selection for estimating sugarcane leaf nitrogen concentration using EO-1 Hyperion hyperspectral data[J]. International Journal of Remote Sensing,2013,34(2):712-728.
[5]黄双萍,洪添胜,岳学军,等.基于高光谱的柑橘叶片氮素含量多元回归分析[J].农业工程学报,2013,29(5):132-138.HUANG Shuangping,HONG Tiansheng,YUE Xuejun,et al. Multiple regression analysis of citrus leaf nitrogen content using hyperspectral technology[J]. Transactions of the CSAE,2013,29(5):132-138.(in Chinese)
[6] TONG Peijin,DU Yiping,ZHENG Kaiyi,et al. Improvement of NIR model by fractional order Savitzky-Golay derivation(FOSGD)coupled with wavelength selection[J]. Chemometrics and Intelligent Laboratory Systems,2015,143:40-48.
[7] MENESATTI P,ANTONUCCI F,PALLOTTINO F,et al. Estimation of plant nutritional status by vis-NIR spectrophotometric analysis on orange leaves[J]. Biosystems Engineering,2010,105(4):448-454.
[8]刘艳丽,何绍兰,吕强,等.柑橘花钾素营养的高光谱表征[J].果树学报,2014,31(6):1065-1071.LIU Yanli,HE Shaolan,LQiang,et al. Characteristics of potassium content in citrus flowers with hyperspectral imagery[J].Journal of Fruit Science,2014,31(6):1065-1071.(in Chinese)
[9]朱西存,姜远茂,赵庚星,等.基于模糊识别的苹果花期冠层钾素含量高光谱估测[J].光谱学与光谱分析,2013,33(4):1023-1027.ZHU Xicun,JIANG Yuanmao,ZHAO Gengxing,et al. Hyperspectral estimation of kalium content in apple florescence canopy based on fuzzy recognition[J]. Spectroscopy and Spectral Analysis,2013,33(4):1023-1027.(in Chinese)
[10]赵化兵,王洁,董彩霞,等.基于可见/近红外反射光谱的梨树叶片钾含量的快速测定研究[J].土壤,2014,46(2):256-261.ZHAO Huabing,WANG Jie,DONG Caixia,et al. Fast determination of potassium content in fresh pear leaves using visible/near infrared spectroscopy technique[J]. Soils,2014,46(2):256-261.(in Chinese)
[11]屈莎,李振海,邱春霞,等.基于开花期氮素营养指标的冬小麦籽粒蛋白质含量遥感预测[J].农业工程学报,2017,33(12):186-193.QU Sha,LI Zhenhai,QIU Chunxia,et al. Remote sensing prediction of winter wheat grain protein content based on nitrogen nutrition index at anthesis stage[J]. Transactions of the CSAE,2017,33(12):186-193.(in Chinese)
[12] PAN S J,YANG Qiang. A survey on transfer learning[J]. IEEE Transactions on Knowledge and Data Engineering,2010,22(10):1345-1359.
[13] BARNETT S M,CECI S J. When and where do we apply what we learn? A taxonomy for far transfer[J]. Psychological Bulletin,2002,128(4):612-637.
[14] LU Jie,BEHBOOD V,HAO Peng,et al. Transfer learning using computational intelligence:a survey[J]. Knowledge-Based Systems,2015,80(Supp.):14-23.
[15]岳学军,全东平,洪添胜,等.不同生长期柑橘叶片磷含量的高光谱预测模型[J].农业工程学报,2015,31(8):207-213.YUE Xuejun,QUAN Dongping,HONG Tiansheng,et al. Prediction model of phosphorus content for citrus leaves during different growth periods based on hyperspectrum[J]. Transactions of the CSAE,2015,31(8):207-213.(in Chinese)
[16]岳学军,全东平,洪添胜,等.柑橘叶片叶绿素含量高光谱无损检测模型[J].农业工程学报,2015,31(1):294-302.YUE Xuejun,QUAN Dongping,HONG Tiansheng,et al. Non-destructive hyperspectral measurement model of chlorophyll content for citrus leaves[J]. Transactions of the CSAE,2015,31(1):294-302.(in Chinese)
[17] MAHAJAN G R,SAHOO R N,PANDEY R N,et al. Using hyperspectral remote sensing techniques to monitor nitrogen,phosphorus,sulphur and potassium in wheat(Triticum aestivum L.)[J]. Precision Agriculture,2014,15(5):499-522.
[18] GOMEZ-CASERO M T, LOPEZ-GRANADOS F, PENA-BARRAGAn J M, et al. Assessing nitrogen and potassium deficiencies in olive orchards through discriminant analysis of hyperspectral data[J]. Journal of the American Society for Horticultural Science,2007,132(5):611-618.
[19] WU Di,WU Hongxi,CAI Jingbo,et al. Classifying the species of exopalaemon by using visible and near infrared spectra with uninformative variable elimination and successive projections algorithm[J]. Journal of Infrared and Millimeter Waves,2009,28(6):423-427.
[20]邓小蕾,李民赞,郑立华,等.基于反射光谱预处理的苹果叶片叶绿素含量预测[J].农业工程学报,2014,30(14):140-147.DENG Xiaolei,LI Minzan,ZHENG Lihua,et al. Estimating chlorophyll content of apple leaves based on preprocessing of reflectance spectra[J]. Transactions of the CSAE,2014,30(14):140-147.(in Chinese)
[21] TAO Chao,PAN Hongbo,LI Yansheng,et al. Unsupervised spectral-spatial feature learning with stacked sparse autoencoder for hyperspectral imagery classification[J]. IEEE Geoscience and Remote Sensing Letters,2015,12(12):2438-2442.
[22] ZABALZA J,REN J,ZHENG Jiangbin,et al. Novel segmented stacked autoencoder for effective dimensionality reduction and feature extraction in hyperspectral imaging[J]. Neurocomputing,2016,185:1-10.
[23] ZHAO Chunhui,WAN Xiaoqing,ZHAO Genping,et al. Spectral-spatial classification of hyperspectral imagery based on stacked sparse autoencoder and random forest[J]. European Journal of Remote Sensing,2017,50(1):47-63.
[24] HINTON G E,SALAKHUTDINOV R R. Reducing the dimensionality of data with neural networks[J]. Science,2006,313(5786):504-507.
[25] LECUN Y,BENGIO Y,HINTON G. Deep learning[J]. Nature,2015,521(7553):436-444.
[26]岳学军,凌康杰,洪添胜,等.基于高光谱图像的龙眼叶片叶绿素含量分布模型[J/OL].农业机械学报,2018,49(8):18-25.YUE Xuejun,LING Kangjie,HONG Tiansheng,et al. Distribution model of chlorophyll content for longan leaves based on hyperspectral imaging technology[J/OL]. Transactions of the Chinese Society for Agricultural Machinery,2018,49(8):18-25. http:∥www. j-csam. org/jcsam/ch/reader/view_abstract. aspx? flag=1&file_no=20180802&journal_id=jcsam. DOI:10. 6041/j. issn. 1000-1298. 2018. 08. 002.(in Chinese)
[27]孙俊,丛孙丽,毛罕平,等.基于高光谱的油麦菜叶片水分CARS-ABC-SVR预测模型[J].农业工程学报,2017,33(5):178-184.SUN Jun,CONG Sunli,MAO Hanping,et al. CARS-ABC-SVR model for predicting leaf moisture of leaf-used lettuce based on hyperspectral[J]. Transactions of the CSAE,2017,33(5):178-184.(in Chinese)
[28] ZHOU Weixun,SHAO Zhenfeng,DIAO Chunyuan,et al. High-resolution remote-sensing imagery retrieval using sparse features by auto-encoder[J]. Remote Sensing Letters,2015,6(10):775-783.
[29]王珂,沈掌泉,ABOU-ISMAIL O,等.不同钾营养水平的水稻冠层和叶片光谱特征研究初报[J].科技通报,1997,13(4):211-214.WANG Ke,SHEN Zhangquan,ABOU-ISMAIL O,et al. Preliminary study on canopy and leaf reflectance characteristics of rice with various potassium levels[J]. Bulletin of Science and Technology,1997,13(4):211-214.(in Chinese)
[30]刘延,吴秋菊,舒清态.基于非成像高光谱的云烟97号叶片钾含量估算研究[J].山东农业科学,2018,50(4):138-141.LIU Yan,WU Qiuju,SHU Qingtai. Estimation study on potassium content in Yunyan 97 leaves based on non-imaging hyperspectral data[J]. Shandong Agricultural Sciences,2018,50(4):138-141.(in Chinese)
[31] ZHONG Ping,WANG Runsheng. Jointly learning the hybrid CRF and MLR model for simultaneous denoising and classification of hyperspectral imagery[J]. IEEE Transactions on Neural Networks and Learning Systems,2014,25(7):1319-1334.
[32]刘炜,常庆瑞,郭曼,等.土壤导数光谱小波去噪与有机质吸收特征提取[J].光谱学与光谱分析,2011,31(1):100-104.LIU Wei,CHANG Qingrui,GUO Man,et al. Extraction of first derivative spectrum features of soil organic matter via wavelet de-noising[J]. Spectroscopy and Spectral Analysis,2011,31(1):100-104.(in Chinese)
[33]梁亮,杨敏华,臧卓.基于小波去噪与SVR的小麦冠层含氮率高光谱测定[J].农业工程学报,2010,26(12):248-253.LIANG Liang,YANG Minhua,ZANG Zhuo. Determination of wheat canopy nitrogen content ratio by hyperspectral technology based on wavelet denoising and support vector regression[J]. Transactions of the CSAE,2010,26(12):248-253.(in Chinese)
[34]张艳超,肖宇钊,庄载椿,等.基于小波分解的油菜多光谱图像与深度图像数据融合方法[J].农业工程学报,2016,32(16):143-150.ZHANG Yanchao,XIAO Yuzhao,ZHUANG Zaichun,et al. Data fusion of multispectral and depth image for rape plant based on wavelet decomposition[J]. Transactions of the CSAE,2016,32(16):143-150.(in Chinese)