【作者】 殷晓平；

【导师】 赵杰文；

【作者基本信息】 江苏大学 ， 农产品加工及贮藏工程， 2010， 硕士

【摘要】 叶绿素是植物叶片的基本组成物质之一,是植物生长和受环境胁迫等情况的敏感指示器。植物在营养素缺乏或者受到其他外界环境干扰时,都会在叶片叶绿素的含量和分布上表现出来。利用新技术快速、准确和无损的测量植物叶片的叶绿素含量及其分布,替代费时费力的化学分析方法,对植物长势检测与估产、营养诊断与施肥等有重要意义。本研究以黄瓜叶片为研究对象,探讨应用高光谱图像技术检测叶片叶绿素含量及其分布的方法,主要的研究内容如下：1按照常规方法提取黄瓜叶片高光谱图像特征参数,如红边参数和植被指数,分析它们与叶绿素含量之间的关系,发现各参数与叶绿素之间存在一定的相关关系,但是相关性均不高。结果表明,红边参数和植被指数反映的信息较单一,而且必须针对具体情况对其修正,有很大的局限性；2研究主成分分析法和独立分量法提取高光谱图像特征参数,提取黄瓜叶片高光谱图像光谱维的前10主成分分量和前8个独立分量,分别利用多元回归建立叶绿素含量的预测模型,其预测集相关系数R分别达到0.827和0.831。利用逐步回归比较两种方法,独立分量分析法只需要一个独立分量既能与叶绿素含量的相关系数R达到0.766,而主成分分析法需要前3个主成分综合才能得到相似效果。结果表明,利用独立分量法分析黄瓜叶片高光谱图像,预测叶片叶绿素含量的方法是可行的,且独立分量分析比主成分分析更有优势；3首次根据独立分量分析法得到的叶绿素含量预测模型,计算出黄瓜叶片叶绿素的分布图。结果表明利用分离出来的独立分量计算得到的黄瓜叶片叶绿素含量分布图与实际情况相符合,为植物营养元素亏缺等研究奠定基础；4用IDL(interface description language)对ENVI (The Environment for Visualizing Images)进行二次开发,开发出了一套高光谱图像数据处理软件,集成了基于批量处理的高光谱图像的标定、感兴趣区域提取、各波长图像及其纹理信息提取、独立分量图计算、数据输出等功能,为快速有效的处理高光谱图像海量数据提供了思路,为科研提供了便利。本论文对利用高光谱图像技术预测黄瓜叶片叶绿素含量及其分布进行了初步研究,探索了独立分量法在高光谱图像处理中的应用。研究成果不仅为高光谱图像处理探索了新的方法,而且为利用高光谱图像技术快速无损检测植物叶片叶绿素含量及其分布提供了充分的理论依据,对实现植物营养诊断、病虫害判断和植物长势监测等方面的自动化、智能化有着积极的意义。更多还原

【Abstract】 Chlorophyll is one of basic components of plant. It is sensitive Indicator of plant environmental stress and growth. When nutrition deficiency or other Interference happened on plant, the concentration and distribution of chlorophyll were changed. As an important matter containing nitrogen in plant, the chlorophyll also indicates the nitrogen absorption and utilization. In stead of the time-consuming, costly chemical analysis method, to develop a new approach to rapidly, accurately and non-destructive detect the concentration and distribution of chlorophyll is meaningful.By taking cucumber leaf as the research object, hyper-spectral imaging was used to detect concentration and distribution of chlorophyll in cucumber leaf in the study. The main research results were as fallows:1 Red edge parameters and vegetation index were extracted and studied, there was some relationship between the parameters and the concentration of chlorophyll, but it was not significant. The result showed that the information reflected from these parameters was not complete. Those parameters had to be corrected in special condition and could not used to determine the concentration of chlorophyll directly.2 10 principal components factors (PCs) and 8 independent component factors (ICs) were calculated by PCA (Principal Component Analysis) and ICA (Independent Component Analysis). Multiple linear regression was applied to build the quantitative model with the PCs and ICs. The correlation coefficient (R) was, respectively,0.827 and 0.831. Stepwise regression was applied to compare the effect of ICA and PCA. The result shows that R is up to 0.766 when only using one IC. ICA was better in detecting the concentration of chlorophyll, and had advantages in separating the signal of chlorophyll in hyper-spectral image.3 Based on the model, distribution map of chlorophyll in cucumber leaf was plotted. The result showed calculating distribution map of chlorophyll in cucumber leaf was feasible, and laid a foundation for the further study.4 Software for hyper-spectral image processing based on redevelopment of ENVI was designed. In accordance with requirement of agricultural products inspection, the system possessed the following functions:batch calibration, region of interest extraction, feature extraction, independent component image calculates and data output. The Software could provide convenience for data process of hyper-spectral image.In this paper, detection of concentration and distribution of chlorophyll in cucumber leaf and ICA approaching on the hyper-spectral image were studied. The result provided sufficient theoretical basis for chlorophyll detection and hyper-spectral image processing. It was meaningful for promoting the automation and intelligent in nutrition, plant insect pests and growth vigor diagnosis of plant.更多还原