基于WSN和单目视觉的作物长势监测应用与研究

发布时间：2018-06-19 03:21

本文选题：无线传感器网络 + 单目视觉　；参考：《内蒙古大学》2017年硕士论文

【摘要】：现代农业是精准化的、信息化的农业,将无线传感器网络(WSN)、机器视觉、图像处理和数据挖掘等应用于作物长势的监测及生长环境与作物长势的相关性分析上具有重要的研究价值。本文首先介绍将这些技术应用到作物长势监测和分析上的研究现状和意义。其次,介绍基于WSN的作物生长环境参数监测系统及其在农田环境中的布置工作,且研究玉米生长期中的七种环境参数(光照强度、土壤温度、土壤湿度、空气温度、空气湿度、空气CO2浓度和土壤含盐量)的变化规律和相关性;再次,基于单目视觉技术提出一种非接触式的作物单株高度和宽度检测方法,分析玉米单株高宽性状与叶片长宽性状的变化,且结合生长环境参数构建玉米单株高宽性状和叶片长宽性状的回归模型,并进行对比分析。然后,本文对作物群体表型长势参量覆盖面积的测量及基于图像平面的检测误差进行研究。实验结果表明,玉米生长期中各生长环境参数的变化存在一定的规律性,且不同参数间存在一定的相关性;基于单目视觉的作物单株高度和宽度检测误差范围分别为0.95%～4.76%和0%～6.73%,作物覆盖面积检测误差范围为0.76%～13.42%,均能够获得理想的检测结果:两类作物长势表型性状与生长环境参数间的拟合判定系数可达0.8以上,基于三维曲面拟合的面积检测误差拟合判定系数可达到0.9758。最后,对本文研究工作的不足及可进一步优化的方法进行总结和展望。
[Abstract]:Modern agriculture is precision, information-based agriculture, wireless sensor networks, WSNs, machine vision, The application of image processing and data mining in crop growth monitoring and correlation analysis between growth environment and crop growth has important research value. This paper first introduces the research status and significance of applying these techniques to crop growth monitoring and analysis. Secondly, the monitoring system of crop growth environment parameters based on WSN and its arrangement in farmland environment are introduced, and seven environmental parameters (light intensity, soil temperature, soil humidity, air temperature, air humidity) in maize growing period are studied. Thirdly, based on monocular vision technology, a non-contact method for measuring plant height and width was proposed to analyze the changes of height and width of maize per plant and leaf length and width. The regression models of height and width per plant and leaf length and width were constructed and compared with the growth environment parameters. Then, the measurement of crop population phenotypic growth parameter covering area and the detection error based on image plane are studied in this paper. The results showed that there was a certain regularity in the variation of growth environment parameters in maize growth period, and there was a certain correlation between different growth environment parameters. The detection error range of plant height and width based on monocular vision is 0.954.76% and 0.6.73, respectively, and the error range of crop coverage detection is 0.760.13.42. The ideal results can be obtained: phenotypic growth of two kinds of crops. The fitting and judging coefficient between the shape and the growth environment parameters can be more than 0.8. The error fitting decision coefficient of area detection based on 3D surface fitting can reach 0.9758. Finally, the shortcomings of this study and the methods that can be further optimized are summarized and prospected.
【学位授予单位】：内蒙古大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：S126;TP212.9;TN929.5

【相似文献】