【作者】 石吉勇；

【导师】 赵杰文；

【作者基本信息】 江苏大学 ， 食品科学， 2012， 博士

【摘要】 随着现代农业的快速发展,我国设施农业种植面积已达到330万公顷,居世界第一。由于设施栽培作物生长周期短、产出量大、需肥量多,从而要求土壤肥力供给具有高消耗、高补充的特点,生产中设施栽培作物很容易出现氮、磷、钾等营养元素亏缺状况,严重影响了作物的产量和经济效益。快速、准确的诊断作物营养元素亏缺对提高作物产量、增加农民收入、发展现代农业具有极其重要的意义。为了克服现有设施栽培作物营养元素亏缺诊断方法的不足,本研究尝试采用高光谱图像技术代替叶片光谱分析技术和计算机图像技术来表征营养元素亏缺引起的作物内部组分信息和外部形态特征,期望得到一种设施栽培作物营养元素亏缺诊断新方法。本文的主要研究工作如下：(1)营养元素亏缺样本培育。在温室大棚中以无土栽培方式培育了氮(N)、磷(P)、钾(K)、镁(Mg)元素亏缺及对照组黄瓜植株；分别采用凯氏定氮法、分光光度计法、原子吸收法分析了缺N组、缺P组、缺K组、缺Mg组和对照组黄瓜植株中N、P、K和Mg元素含量的分布规律。结果表明：N、P、K和Mg元素含量在黄瓜植株新叶、中叶、老叶中呈递减趋势；与对照组叶片相比较,缺N组、缺P组、缺K组和缺Mg组植株老叶的N、P、K、Mg元素含量下降明显,说明缺N组、缺P组、缺K组和缺Mg组中的黄瓜植株进入了各自对应的营养元素亏缺状态,营养元素亏缺植株的成功培育为后续研究奠定了坚实的基础。(2)样本缺素症状及其分布规律初步分析。采集缺N组、缺P组、缺K组、缺Mg组和对照组黄瓜植株每个节点上的黄瓜叶片作为分析样本,采用高效液相色谱技术(HPLC)检测样本的叶绿素、叶黄素和β胡萝卜素含量,采用彩色相机拍摄样本的图像信息,通过分析N、P、K和Mg元素亏缺样本的色素含量变化和外观特征,研究了不同营养元素亏缺症状及其在作物植株及作物叶片中的分布规律。结果表明：N元素亏缺植株的老叶叶绿素含量降低,叶面整体退绿；P元素亏缺植株老叶叶绿素没有明显变化,但是在老叶叶脉附近出现水渍斑；K元素亏缺植株老叶叶缘处的叶绿素含量降低；Mg元素亏缺植株老叶叶脉间的叶绿素含量降低；缺素症状及其分布规律的分析结果为解决如何采集具有代表性的缺素样本、采用什么技术在样本的哪些区域提取何种缺素特征等元素亏缺诊断中的关键问题指明了方向。(3)叶绿素含量叶面分布图检测。提取120片黄瓜叶片的可见-近红外高光谱图像光谱信息,采用区间偏最小二乘法(iPLS)、联合区间偏最小二乘法(SiPLS)、遗传算法-区间偏最小二乘法(GA-iPLS)、遗传算法-模拟退火算法-区间偏最小二乘法(GA-SA-iPLS)优选了与叶绿素含量相关的特征区间(分别为第10、11、13、17、18区间),利用主成分分析(PCA)和独立分量分析(ICA)提取了入选区间的光谱特征,并将光谱特征与HPLC检测得到的叶绿素含量值进行关联,建立了叶片高光谱图像信号与叶绿素含量之间的对应关系(Rp=0.8769, RMSEP=2.42mg/g);依次将高光谱图像中每个像素点的光谱代入叶绿素含量回归模型中计算出每个像素点对应的叶绿素含量值,得到了叶片叶绿素含量的叶面分布图。该叶绿素含量分布图解决了化学分析方法、光谱分析法难以检测整个叶面上叶绿素含量分布的难题。(4)基于叶绿素含量叶面分布的N、K和Mg元素亏缺诊断。检测了对照组叶片、N元素亏缺叶片、K元素亏缺叶片和Mg元素亏缺叶片的叶绿素含量叶面分布图；根据叶绿素含量叶面分布规律提出了N、K和Mg元素亏缺诊断新方法,对N、K和Mg元素亏缺诊断率分别为97%、90%、90%。新方法克服了化学诊断方法速度慢、破坏检测对象的缺点,弥补了光谱诊断法重现性差、准确度低及信息获取不全面的不足。(5)基于近红外高光谱图像的P元素亏缺诊断。采集了对照组叶片、P元素亏缺叶片的近红外高光谱图像,分别采用PCA和ICA提取了近红外高光谱图像的特征图像；结果表明第1独立分量图像能够有效的表征P元素亏缺引起的缺素症状,同时近红外高光谱图像结合ICA能够提取肉眼不可见的轻微P元素亏缺症状。采用该方法对30片P元素亏缺叶片和30片对照组叶片进行了诊断,诊断率为98%。(6)设施栽培作物营养元素亏缺的数据库软件开发。该软件能够详细记录作物的种类、品种、元素亏缺种类、元素亏缺症状图片及光谱等信息。为判别作物缺素情况的数字化信息描述做了开创性的工作。本研究提供了新的设施栽培作物营养元素亏缺无损诊断方法,研究成果对提高我国设施栽培技术水平有着积极的意义。更多还原

【Abstract】 With the rapid development of modern agriculture, China has the world’s largest area of facility agriculture, with a total of 3.3 million ha. As the establishment planting crop’s short growth cycle, large yield and fertilizer requirements, leads to high-cost and fast-replenish in soil fertility. This used to cause nitrogen (N), phosphorus (P) and magnesium (K) deficiencies in the growth of cultivated crops, affect the yield and economic profits. Fast and accurate diagnostics of nutrient deficiency in cultivated crops has very important significance of increasing yield and the income of framers, the development of modern agriculture. In order to overcome the disadvantages of existing methods for diagnostics of nutrient deficiency in cultivated crops, hyper-spectral imaging technology was used to characterize physiological and morphological symptoms resulting from nutrient deficiency. We hope to obtain a new method for diagnostics of nutrient deficiency in cultivated crops. The main research works in this paper has been summed as following:(1) Cultivation of nutrient deficient samples. N, P, K, Mg deficient mini-cucumber plants and control plants were grown under non-soil conditions. Kjeldahl method, spectrophotometer method and atomic absorption spectrometry were used to determine N, P, K and Mg concentrations of mini-cucumber plants in N, P, K, Mg deficient and control groups. Results showed that:N, P, K and Mg concentrations for new, middle and old leaves were in a decreasing order. In comparison with N, P, K and Mg concentration of older leaves in control group, N, P, K and Mg concentration of older leaves in nutrient deficient groups decline markedly. This indicated plants in N, P, K and Mg deficient groups entered into nutrient deficient states. The successful cultivation of nutrient deficient samples laid a good foundation for further analysis.(2) Analysis of nutrient deficient symptoms and its distribution. Leaves at every single node in plant of N, P, K, Mg deficient and control groups were used as analytical samples. Chlorophyll, lutein and carotene concentrations of leaf sample were determined by high performance liquid chromatography (HPLC); image information was recorded by a color camera, then the nutrient deficient symptoms and its distribution in whole plant and leaf were analyzed. Results showed that:N deficiency caused to chlorophyll decreased in the whole older leaves, P deficiency did not change the leaf chlorophyll content, but resulting in the occurrence of small chlorotic spots, K deficiency caused to chlorophyll decreased in edge of older leaves, Mg deficiency caused to chlorophyll decreased between vein of older leaves. The analytical result figured out how to collect analytical sample, which technology to be used, and what kind of deficient symptoms to be extracted for further analysis.(3) Determination of chlorophyll concentration distribution map on mini-cucumber leaf. Spectra of 120 cucumber hyper-spectral images were extracted; interval partial least-squares (iPLS), synergy interval partial least-squares (siPLS), genetic algorithm interval partial least-squares (GA-iPLS) and genetic algorithm- simulated annealing algorithm-interval partial least-squares (GA-SA-iPLS) were used to select the efficient wavelength regions for chlorophyll concentrations (No.10,11,13,17,18), Principal Component Analysis (PCA) and Independent Component Analysis (ICA) were used to extract information from the selected wavelength regions, then the optimal chlorophyll calibration model was obtained (Rp=0.8769, RMSEP=2.42mg/g), spectral of every pixel in hyper-spectral images were computed and chlorophyll content of every pixel was obtained according to the calibration model. Finally, the chlorophyll distribution map was estimated. In comparison with chemical and spectral methods, the hyper-spectral technology could determine the chlorophyll concentration in the whole leaf.(4) Diagnostics of N, K and Mg deficiencies base on chlorophyll concentration distribution map. Chlorophyll concentration distribution map of N, K, Mg deficient and control leaves were determined. According to their chlorophyll concentration distribution maps, new diagnostics methods for N, K, and Mg deficiencies were proposed, the diagnostic rate for N, K, and Mg deficiencies were 97%,90% and 90%; respectively. The hyper-spectral diagnostic method overcomes the disadvantages of chemical and spectral diagnostic methods.(5) Diagnostics of P deficiency base on near infrared hyper-spectral image. Leave in P deficient and control group were used to collect near infrared hyper-spectral image; PCA and ICA were used to extract characteristic images. Results showed that the first ICA image could be used to extract P deficient symptoms; furthermore, this method could diagnose P deficient symptoms prior to the occurrence of responses to P deficiency that can be observed visually.(6) Development of nutrient deficient database soft for cultivated crops. This soft could record the kind of crop, varieties, kind of deficient nutrient, deficient image and spectra. It could be used for digital description of symptoms caused by nutrient deficiencies.In this study, new methods for non-destrctive diagnostics of nutrient elements deficiencies are proposed, and there is also of great significant in improve the level of facility agriculture in China.更多还原