【作者】 陈丽君；

【导师】 任文涛；

【作者基本信息】 沈阳农业大学 ， 农业机械化工程， 2009， 博士

【摘要】 普通植物保护机械不考虑杂草和作物在空间上的区域分布特征,将农药均匀地喷洒在田间或整个行上,不仅造成了农药的大量浪费、污染了生态环境,严重时还影响到农作物的生长,最终会影响产量。国内外研究表明,实施精准、精量喷雾,可节约农药40%-60%。本文在对国内外相关文献深入分析评述的基础上,以穴播玉米幼苗为作业对象,针对穴播作物的位置特征,设计了基于机器视觉的变量喷雾控制系统,构建了该控制系统的软、硬件试验平台。在实验室内进行了动态模拟试验,对各算法进行了调整并对系统的匹配模型进行了验证。本文的主要研究内容如下:(1)对运动图像的实时采集方法进行了研究。分析了各采集设备的合理配置,以及自然光照条件下获取清晰运动图像的方法,实现了动态图像的实时采集,为进一步的田间图像的实时处理研究作了准备。(2)利用自动取阈法对绿色植物与背景进行了分割。在彩色图像分割中,通过对比分析研究了绿色作物与背景的各种分割方法,确定了超绿色灰度化方法,并利用Ostu最大类间方差自动取阈法对绿色植物与背景进行了分割。试验结果表明,此算法具有很好的适应性,对动态图像的分割效果很好。(3)研究了作物株心定位方法。对株心定位方法进行了比较研究,针对田间穴播作物按规律分布生长的特点和田间场景中植物叶片交叠等问题,提出了一种基于作物位置特征的过分割区块投影定位作物株心的方法。通过对灰度化图像进行过度分割,滤出大部分杂草像素,在过分割图像中进行中心区块投影确定作物株心,提高了定位精度。依据获得的株心位置,划分喷雾区块进行像素的统计判别,最终生成相应的喷雾控制指令编码。(4)设计了利用计算机并口控制电磁阀的驱动系统。分析了当前国内大多数研究都采用主从式控制方式控制电磁阀方法的固有缺点,提出了基于并口的控制方法。省去下位机,直接采用主机闲置的并口资源对电磁阀进行控制。实验证明,该方法能很好地实现对电磁阀的控制。(5)确定了株心坐标的透视投影调整模型。为了确定图像株心坐标与世界绝对坐标之间的关系,以便精确控制喷雾位置,依据摄像机透视成像原理推导出了株心坐标的透视投影调整和坐标转换的数学模型,并且通过试验验证了模型的准确性。(6)确定了穴间变量喷雾控制系统的匹配模型,构建了基于机器视觉的穴间变量喷雾控制系统试验平台。系统中各部分的匹配协调,是喷雾是否准确的关键。针对硬件系统中电磁阀、压力系统及喷头等的机械特性,实验测定了本系统的动作延迟参数,建立了穴间变量喷雾控制系统的匹配模型,并对各主要匹配参数进行了分析与设计,试验验证了系统模型的准确性。采用VB与Matlab混合编程的方法编写了基于机器视觉的变量喷雾控制系统试验软件。研究成果为变量喷雾机的研制奠定了基础。更多还原

【Abstract】 Herbicides and pesticides are spayed evenly in the whole fields or crops lines by general plant protection machinery, not considering the regional distribution of weeds and crops in space, which not only resulted in a large number of pesticide waste, but also pollution of the ecological environment and influence in the growth and yield of crops. Domestic and international studies have shown that it can save 40% -60% of pesticides if accurate application at the crop or weeds is used. At the foundation of summarized the domestic and foreign related research, this article studied an inter-variable spray control system which is based on the machine vision, also designed the software and hardware test platform for block spray control, determined the image processing algorithms and adjust various algorithms under the simulation in the laboratory. Ultimately the spray point between the variables is realized. The main research content of this article is as follows:(1) The methods of acquiring real-time motion image have been studied, and the rational allocation of collection equipment and the methods of obtaining a clear image of movement under light conditions have been analyzed. The real-time dynamic image acquisition prepared for further study of real-time processing of images in the fields.(2) The background of green plants has been divided through automatic threshold method. By comparing a variety of methods, it ultimately adopted the ultra-green method of image gray level, using Otsu automatic threshold method to detach green plants and the background of the partition. Tests showed that this algorithm has good adaptability, and has a good segmentation for the dynamic image.(3) Methods of positioning the center of the crop strains have been studied. As the crops are bunch planting and are regularly planted at uniform intervals along the line of growth, this paper presents the location of the characteristics of crops based on the over-segmentation positioning block projection method to locate the heart of crops. That is, on the gray-scale images of over-segmentation, most weeds are filtered out. The heart was determined in the over-segmentation, so that the positioning accuracy was greatly improved. Pixel blocks divided spray statistical discrimination according to the located crop heart, and ultimately to generate the corresponding control code spray.(4) A drive circuit using computer parallel port to control the solenoid valve was designed. After analyzing the current studies which adapted master-slave method to control solenoid valve, we proposed a control method based on the parallel port. Saving the Under-bit machine, we directly used the host parallel port resources idle to control solenoid valve. Experimental results show that the method can achieve good control of solenoid valve.(5) In order to precisely control the implementation of the spray hole, we need to determine relationship between the absolute coordinates with the world and that of the heart image the According to the principle of the camera lens imaging, we derived conversion formula between the coordinates of crop heart and coordinate, and its accuracy was verified by experiments.(6) A model that matches the Inter-hole variable spray control system was built. A machine vision-based inter-hole variable spray control system test platform has been build. The coordination of the various parts of the match in the system is the key to the accuracy of spraying. According to the features of hardware system for solenoid valves, pressure sprinkler systems and the mechanical properties, the experiment determined delay time of the system, integrated image acquisition and processing and finally derived the Inter-hole variable spray control system. It was certificated that it can meet the test of real-time processing requirements and the applied software. The use of VB and Matlab mixed programming method prepared machine vision-based inter-hole test variable spray control system software, laying the foundation for the further development of sprayer.更多还原