【作者】 孙棋；

【导师】 王珂；

【作者基本信息】 浙江大学 ， 农业遥感与信息技术， 2008， 硕士

【摘要】 本文选择田间水稻为研究对象,利用旋翼无人机获取不同氮素水平下的水稻冠层图像,同时在室内获取水稻样品叶片扫描图像,通过数字图像处理技术建立水稻氮素水平诊断指标,得到水稻氮素营养诊断的初步研究结果。具体如下:(1)本研究采用日本新一代旋翼式无人航空摄影平台HeraklesⅡ,自带高稳定性减震拍摄系统。选取日本Canon公司的EOS 30D型数码单反(单镜头反光)相机作为传感器。从摄影系统采集的数据看,本次实验的影像质量较高,田块边界比较清晰,每组影像的亮度、景深等都有所差异。这样在后期的影像信息提取工作中有更多样化的数据可供选择;(2)选用扫描仪作为传感器获取水稻叶片的数字图像,通过水稻氮素和叶绿素含量、SPAD值之间的相关性分析,得到有效的颜色特征参量B、b、b/(r+g)、b/r、b/g。同时比较不同叶位、不同位点的变异系数,选择较为稳定的第三完全展开叶(L2)作为指示叶或参照叶;最后建立不同氮素水平的识别模型,得到四个氮素水平的正确识别率为:N0:74.9%;N1:52%;N2:84.7%;N3:75%;(3)通过对旋翼无人机获取的水稻图像进行色彩分析,结果表明,无人机拍摄冠层图像的颜色特征参量G值与水稻叶片的SPAD、叶绿素和氮含量均有很好的相关关系。通过高光谱遥感数据的相关分析,从机理上说明基于数字图像处理技术的水稻氮素营养诊断是可行的。引入深绿色指数DGCI,研究认为颜色特征参量G值和DGCI可以用来表征水稻拔节期氮素营养状况。同时建立不同氮素水平的水稻冠层数字图像识别模型,得到四个氮素水平的正确识别率为:N0:91.6%;N1:70.83%;N2:86.7%;N3:95%。更多还原

【Abstract】 In this paper, the remotely-controlled helicopter (Herakles II) was selected as the monitoring platform and digital camera (EOS 30D) was used to collect canopy data of rice. And the scanner was adopted as the digital image sensor to collect information of leaf, the method and model of diagnosing the status of rice was established based on image processing.(1) In this study, the remotely-controlled helicopter (Herakles II) was selected as the monitoring platform and digital camera (EOS 30D) was used to collect data.These two are both produced from Japan. The digital image got from the remotely-controlled helicopter was quite well, the boundary between different fields was clear and the variety of light was obvious. It would provide enough datas to extract information for later studying from these images.(2) According to the analysis of relations between value of rice leaf SPAD, leaf percentage nitrogen contents and plant percentage nitrogen contents, the effective color parameter had been abstraced as B、b、b/(r+g)、b/r、b/g. It was suggested that the third leaf from the top was the most ideal indicator and it was choosen as the object researched. Finally the method and model of diagnosing the status of rice was established and the accuracy were as follows: NO: 74.9%; N1: 52%; N2: 84.7%; N3: 75%.(3) It is suggested that there was a good relation between the data of G, value of rice leaf SPAD, leaf percentage nitrogen contents and plant percentage nitrogen contents. The correlative relationship between leaf spectral characteristics and hyperspectral remote sensing was studied. It was indicated that the diagnose of nitrogen at different leaf positon based on the color characteristic values is feasible. The DGCI was introduced to certify that the data of G and DGCI could express the different status of rice on the jointing stage. Finally the method and model of diagnosing the status of rice was established and the accuracy were as follows: NO: 91.6%; N1: 70.83%; N2: 86.7%; N3: 95%.更多还原