基于深度学习的龙眼叶片叶绿素含量预测的高光谱反演模型
|
摘要
【目的】探讨龙眼Dimocarpus longan Lour.叶片发育过程中叶绿素含量二维分布变化规律,实现无损检测病虫害对叶片叶绿素含量分布的影响,为评估嫩叶抗寒能力、龙眼结果期的施肥量和老熟叶的修剪提供参考。【方法】利用高光谱成像仪采集龙眼叶片在369~988 nm区间的高光谱图像,自动提取感兴趣区域,利用分光光度法测定叶片叶绿素含量。基于皮尔森相关系数(r)分析了龙眼叶片生长过程中各波段光谱响应与叶绿素含量之间相关性,建立偏最小二乘回归模型。分析了特征波段图像纹理特征与叶绿素含量相关性,将光谱特征和纹理特征结合导入深度学习中的稀疏自编码(SAE)模型预测龙眼叶片叶绿素含量,结合"图谱信息"的SAE模型预测龙眼叶片叶绿素含量的分布情况。【结果】龙眼叶片3个生长发育期相关系数的曲线均在700 nm附近出现波峰,嫩叶、成熟叶和老熟叶3个阶段相关性最高的波长分别为692、698和705 nm;全发育期的最敏感波段相关性远高于3个生长发育期,r达到0.890 3。回归模型中,吸收带最小反射率位置和吸收带反射率总和建立的最小二乘回归模型预测效果最好(R_c~2=0.856 8,RMSEc=0.219 5;R_v~2=0.771 2,RMSEv=0.286 2),其校正集和验证集的决定系数均高于单一参数建立的预测模型。在所有预测模型中,结合"图谱信息"的SAE模型预测效果最好(R_c~2=0.979 6,RMSEc=0.171 2;R_v~2=0.911 2,RMSEv=0.211 5),且预测性能受叶片成熟度影响相对较小,3个生长阶段R_v~2的标准偏差仅为最小二乘回归模型标准偏差的29.9%。【结论】提出了一种自动提取感兴趣区域的方法,成功率为100%。基于光谱特征的回归模型对不同生长阶段的叶片预测效果变化较大,而基于"图谱信息"融合的SAE模型预测性能受叶片成熟度影响相对较小且预测精度较高,SAE模型适用于不同成熟度的龙眼叶片叶绿素含量分布预测。
【Objective】To study the distribution of chlorophyll content of Longan(Dimocarpus longan Lour)leaves in different growth periods, realize non-destructive measurement of the influence of pests and diseases on chlorophyll distribution, and provide a reference for evaluating the cold-resistant ability of young leaves,fertilizing amount in the fruiting period and pruning of mature leaves.【Method】Hyperspectral images of Longan leaves in three growth periods were acquired via an online hyperspectral imaging system within the spectral region of 369–988 nm wavelength. An automatic masking method was used to extract the interest regions.The chlorophyll content was measured by the spectrophotometric method. The relationships between the spectral response characteristics and chlorophyll contents of Longan leaves in three growth periods were measured based on Pearson correlation coefficient(r). A partial least squares regression(LSR) model was established. The relationship between the texture feature of selected image and chlorophyll content was analyzed. The spectroscopy and texture features were imported to the spare auto-encoder(SAE) model in deep learning to predict the chlorophyll content of Longan leaves. The distribution of chlorophyll content was predicted using SAE model based on the mapping information.【Result】The peaks of correlation coefficient curves of Longan leaves in three growth periods appeared in the vicinity of 700 nm. The wavelength of the highest correlation coefficient for young, mature and old ripe leaves was 692, 698 and 705 nm, respectively. The correlation coefficient(r) of the most sensitive band in full period was higher than those in three growth periods, which was up to 0.890 3. Among all regression models, the prediction effect of LSR model based on the absorption band of the minimum reflectivity and total reflectivity was the best(R_c~2=0.856 8, RMSEc=0.219 5; R_v~2=0.771 2,RMSEv=0.286 2), and the determination coefficients of its calibration and validation sets were higher than those based on a single parameter. SAE model importing spectroscopy and texture features performed the best(R_c~2=0.979 6, RMSEc=0.171 2; R_v~2=0.911 2, RMSEv=0.211 5) and the most stable to predict chlorophyll contents of Longan leaves in different growth periods, its standard deviation was only 29.9% of LSR model.【Conclusion】A method automatically extracting interest region was proposed, its success rate was 100%. The performance of SAE model based on spectroscopy and texture features was more stable than those of regression models based on spectroscopy to predict chlorophyll contents of Longan leaves in different growth periods. SAE model is suitable for predicting the distribution of chlorophyll content of Longan leaves as a non-destructive method.
【Objective】To study the distribution of chlorophyll content of Longan(Dimocarpus longan Lour)leaves in different growth periods, realize non-destructive measurement of the influence of pests and diseases on chlorophyll distribution, and provide a reference for evaluating the cold-resistant ability of young leaves,fertilizing amount in the fruiting period and pruning of mature leaves.【Method】Hyperspectral images of Longan leaves in three growth periods were acquired via an online hyperspectral imaging system within the spectral region of 369–988 nm wavelength. An automatic masking method was used to extract the interest regions.The chlorophyll content was measured by the spectrophotometric method. The relationships between the spectral response characteristics and chlorophyll contents of Longan leaves in three growth periods were measured based on Pearson correlation coefficient(r). A partial least squares regression(LSR) model was established. The relationship between the texture feature of selected image and chlorophyll content was analyzed. The spectroscopy and texture features were imported to the spare auto-encoder(SAE) model in deep learning to predict the chlorophyll content of Longan leaves. The distribution of chlorophyll content was predicted using SAE model based on the mapping information.【Result】The peaks of correlation coefficient curves of Longan leaves in three growth periods appeared in the vicinity of 700 nm. The wavelength of the highest correlation coefficient for young, mature and old ripe leaves was 692, 698 and 705 nm, respectively. The correlation coefficient(r) of the most sensitive band in full period was higher than those in three growth periods, which was up to 0.890 3. Among all regression models, the prediction effect of LSR model based on the absorption band of the minimum reflectivity and total reflectivity was the best(R_c~2=0.856 8, RMSEc=0.219 5; R_v~2=0.771 2,RMSEv=0.286 2), and the determination coefficients of its calibration and validation sets were higher than those based on a single parameter. SAE model importing spectroscopy and texture features performed the best(R_c~2=0.979 6, RMSEc=0.171 2; R_v~2=0.911 2, RMSEv=0.211 5) and the most stable to predict chlorophyll contents of Longan leaves in different growth periods, its standard deviation was only 29.9% of LSR model.【Conclusion】A method automatically extracting interest region was proposed, its success rate was 100%. The performance of SAE model based on spectroscopy and texture features was more stable than those of regression models based on spectroscopy to predict chlorophyll contents of Longan leaves in different growth periods. SAE model is suitable for predicting the distribution of chlorophyll content of Longan leaves as a non-destructive method.
引文
[1]李粉玲,王力,刘京,等.基于高分一号卫星数据的冬小麦叶片SPAD值遥感估算[J].农业机械学报,2015,46(9):273-281.
[2]何启平,陈莹.校园常见植物叶绿素提取方法比较及其含量测定[J].黑龙江农业科学,2015(10):117-120.
[3]丁永军,李民赞,安登奎,等.基于光谱特征参数的温室番茄叶片叶绿素含量检测[J].农业工程学报,2011,27(5):244-247.
[4]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.
[5]李媛媛,常庆瑞,刘秀英,等.基于高光谱和BP神经网络的玉米叶片SPAD值遥感估算[J].农业工程学报,2016,32(16):135-142.
[6]冯雷,骆一凡,何勇,等.基于机器视觉技术的尖椒冠层SPAD值测定仪的开发[J].农业工程学报,2016,32(21):177-182.
[7]赵艳茹,余克强,李晓丽,等.基于高光谱成像的南瓜叶片叶绿素分布可视化研究[J].光谱学与光谱分析,2014,34(5):1378-1382.
[8]余克强,赵艳茹,李晓丽,等.高光谱成像技术的不同叶位尖椒叶片氮素分布可视化研究[J].光谱学与光谱分析,2015,35(3):746-750.
[9]CHENG J H,SUN D W,PU H,et al.Comparison of visible and long-wave near-infrared hyperspectral imaging for colour measurement of grass carp[J].Innov Food Sci Emerg,2014,7(11):3109-3120.
[10]朱瑶迪,邹小波,石吉勇,等.高光谱图像技术快速预测发酵醋醅总酸分布[J].农业工程学报,2014,30(16):320-327.
[11]李芳,石吉勇,张德涛,等.银杏叶中黄酮含量的叶面分布检测研究[J].食品工业科技,2015,36(9):270-272.
[12]JIN H L,LI L L,CHENG J H.Rapid and non-destructive determination of moisture content of peanut kernels using hyperspectral imaging technique[J].Food Anal Method,2015,8(10):2524-2532.
[13]刘顺枝.不同成熟度龙眼叶片组织显微结构的观察[J].嘉应大学学报,2000,18(6):85-86.
[14]黄玉溢,谭宏伟,周柳强,等.龙眼施用长效氮肥的效应研究[J].南方农业学报,2003(S2):71-73.
[15]周庆贤,周颂棠,伦演鹏,等.广州地区石硖龙眼高产栽培技术[J].广东农业科学,2005,32(3):83-84.
[16]彭庆芳.浅谈龙眼优质高产栽培技术[J].吉林农业(学术版),2010(7):107.
[17]潘玲玲,徐晓洁,谭晶晶,等.分光光度法快速测定玉米叶片中的叶绿素[J].分析化学,2007,35(3):413-415.
[18]WU D,SHI H,WANG S,et al.Rapid prediction of moisture content of dehydrated prawns using online hyperspectral imaging system[J].Analytica Chimica Acta,2012,726(9):57-66.
[19]WU D,SUN D W,HE Y.Application of long-wave near infrared hyperspectral imaging for measurement of color distribution in salmon fillet[J].Innov Food Sci Emerg,2012,16(39):361-372.
[20]张永贺,郭啸川,褚武道,等.基于红边位置的木荷叶片叶绿素含量估测模型研究[J].红外与激光工程,2013,42(3):798-804.
[21]薛利红,杨林章.采用不同红边位置提取技术估测蔬菜叶绿素含量的比较研究[J].农业工程学报,2008,24(9):165-169.
[22]梁爽,赵庚星,朱西存.苹果树叶片叶绿素含量高光谱估测模型研究[J].光谱学与光谱分析,2012,32(5):1367-1370.
[23]岳有军,杨雪,赵辉,等.基于支持向量机的油菜缺素诊断研究[J].广东农业科学,2015,42(20):145-148.
[24]尹征,唐春晖,张轩雄.基于改进型稀疏自动编码器的图像识别[J].电子科技,2016,29(1):124-127.
[25]孙文珺,邵思羽,严如强.基于稀疏自动编码深度神经网络的感应电动机故障诊断[J].机械工程学报,2016,52(9):65-71.
[26]LUO W,YANG J,XU W,et al.Locality-constrained sparse auto-encoder for image classification[J].IEEESignal Proc Let,2015,22(8):1070-1073.
[27]VINCENT P,LAROCHELLE H,LAJOIE I,et al.Stacked denoising auto-encoders[J].J Mach Learn Res,2010,11(12):3371-3408.
[28]刘文雅,潘洁.基于神经网络的马尾松叶绿素含量高光谱估算模型[J].应用生态学报,2017,28(4):1128-1136.
[29]邹小波,张小磊,石吉勇,等.基于高光谱图像的黄瓜叶片叶绿素含量分布检测[J].农业工程学报,2014,30(13):169-175.
[30]陈晨,张永成.马铃薯不同品种间气孔密度及叶绿素含量的差异性研究[J].中国农学通报,2013,29(27):83-87.
[31]顾振芳,王伟清,朱爱萍,等.黄瓜对霜霉病的抗性与叶绿素含量、气孔密度的相关性[J].上海交通大学学报(农业科学版),2004,22(4):381-384.
[32]颜惠霞,徐秉良,梁巧兰,等.南瓜品种对白粉病的抗病性与叶绿素含量和气孔密度的相关性[J].植物保护,2009,35(1):79-81.
[2]何启平,陈莹.校园常见植物叶绿素提取方法比较及其含量测定[J].黑龙江农业科学,2015(10):117-120.
[3]丁永军,李民赞,安登奎,等.基于光谱特征参数的温室番茄叶片叶绿素含量检测[J].农业工程学报,2011,27(5):244-247.
[4]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.
[5]李媛媛,常庆瑞,刘秀英,等.基于高光谱和BP神经网络的玉米叶片SPAD值遥感估算[J].农业工程学报,2016,32(16):135-142.
[6]冯雷,骆一凡,何勇,等.基于机器视觉技术的尖椒冠层SPAD值测定仪的开发[J].农业工程学报,2016,32(21):177-182.
[7]赵艳茹,余克强,李晓丽,等.基于高光谱成像的南瓜叶片叶绿素分布可视化研究[J].光谱学与光谱分析,2014,34(5):1378-1382.
[8]余克强,赵艳茹,李晓丽,等.高光谱成像技术的不同叶位尖椒叶片氮素分布可视化研究[J].光谱学与光谱分析,2015,35(3):746-750.
[9]CHENG J H,SUN D W,PU H,et al.Comparison of visible and long-wave near-infrared hyperspectral imaging for colour measurement of grass carp[J].Innov Food Sci Emerg,2014,7(11):3109-3120.
[10]朱瑶迪,邹小波,石吉勇,等.高光谱图像技术快速预测发酵醋醅总酸分布[J].农业工程学报,2014,30(16):320-327.
[11]李芳,石吉勇,张德涛,等.银杏叶中黄酮含量的叶面分布检测研究[J].食品工业科技,2015,36(9):270-272.
[12]JIN H L,LI L L,CHENG J H.Rapid and non-destructive determination of moisture content of peanut kernels using hyperspectral imaging technique[J].Food Anal Method,2015,8(10):2524-2532.
[13]刘顺枝.不同成熟度龙眼叶片组织显微结构的观察[J].嘉应大学学报,2000,18(6):85-86.
[14]黄玉溢,谭宏伟,周柳强,等.龙眼施用长效氮肥的效应研究[J].南方农业学报,2003(S2):71-73.
[15]周庆贤,周颂棠,伦演鹏,等.广州地区石硖龙眼高产栽培技术[J].广东农业科学,2005,32(3):83-84.
[16]彭庆芳.浅谈龙眼优质高产栽培技术[J].吉林农业(学术版),2010(7):107.
[17]潘玲玲,徐晓洁,谭晶晶,等.分光光度法快速测定玉米叶片中的叶绿素[J].分析化学,2007,35(3):413-415.
[18]WU D,SHI H,WANG S,et al.Rapid prediction of moisture content of dehydrated prawns using online hyperspectral imaging system[J].Analytica Chimica Acta,2012,726(9):57-66.
[19]WU D,SUN D W,HE Y.Application of long-wave near infrared hyperspectral imaging for measurement of color distribution in salmon fillet[J].Innov Food Sci Emerg,2012,16(39):361-372.
[20]张永贺,郭啸川,褚武道,等.基于红边位置的木荷叶片叶绿素含量估测模型研究[J].红外与激光工程,2013,42(3):798-804.
[21]薛利红,杨林章.采用不同红边位置提取技术估测蔬菜叶绿素含量的比较研究[J].农业工程学报,2008,24(9):165-169.
[22]梁爽,赵庚星,朱西存.苹果树叶片叶绿素含量高光谱估测模型研究[J].光谱学与光谱分析,2012,32(5):1367-1370.
[23]岳有军,杨雪,赵辉,等.基于支持向量机的油菜缺素诊断研究[J].广东农业科学,2015,42(20):145-148.
[24]尹征,唐春晖,张轩雄.基于改进型稀疏自动编码器的图像识别[J].电子科技,2016,29(1):124-127.
[25]孙文珺,邵思羽,严如强.基于稀疏自动编码深度神经网络的感应电动机故障诊断[J].机械工程学报,2016,52(9):65-71.
[26]LUO W,YANG J,XU W,et al.Locality-constrained sparse auto-encoder for image classification[J].IEEESignal Proc Let,2015,22(8):1070-1073.
[27]VINCENT P,LAROCHELLE H,LAJOIE I,et al.Stacked denoising auto-encoders[J].J Mach Learn Res,2010,11(12):3371-3408.
[28]刘文雅,潘洁.基于神经网络的马尾松叶绿素含量高光谱估算模型[J].应用生态学报,2017,28(4):1128-1136.
[29]邹小波,张小磊,石吉勇,等.基于高光谱图像的黄瓜叶片叶绿素含量分布检测[J].农业工程学报,2014,30(13):169-175.
[30]陈晨,张永成.马铃薯不同品种间气孔密度及叶绿素含量的差异性研究[J].中国农学通报,2013,29(27):83-87.
[31]顾振芳,王伟清,朱爱萍,等.黄瓜对霜霉病的抗性与叶绿素含量、气孔密度的相关性[J].上海交通大学学报(农业科学版),2004,22(4):381-384.
[32]颜惠霞,徐秉良,梁巧兰,等.南瓜品种对白粉病的抗病性与叶绿素含量和气孔密度的相关性[J].植物保护,2009,35(1):79-81.