基于机器视觉技术的作物冠层养分测量仪的研发

发布时间：2018-11-09 11:44

【摘要】：化肥的过度施用,不仅造成了大量的浪费,提高了农业的生产成本,还造成了大量的环境污染和水土流失,非常不利于农业的可持续发展。为了解决这一问题,就需要进入精细农业的思想,对大田里的农作物实现区别对待,按需施肥。要实现这一目标首先必须要实现田间的作物氮素水平快速无损检测。基于这一需求,本研究开发出了一种利用机器视觉技术能够对田间作物冠层的SPAD值进行快速检测的仪器。具体的研究内容和结论如下:(1)利用作物冠层在近红外波段的反射率要远远高于土壤背景在近红外波段的反射率这一特性,制作了一块参照板,使得板上参照色块的近红外反射率位于土壤和冠层之间。将作物冠层在近红外波段下的图像灰度化之后,将参照板的灰度值作为阈值,很好的将作物冠层和土壤背景进行了分离。(2)根据物体的反射率和图像中表示物体像素的灰度值和获取图像时环境光的强度都成一次线性正相关关系这一原理。在参照板上用四种不同的颜色拉开了一个反射率的梯度。并且利用这一梯度建立了反射率和灰度值之间线性模型。成功消除了环境光对作物冠层反射率计算的影响。(3)利用建立的反射率和灰度值动态线性模型,通过求出图像中作物冠层的平均灰度值从而求得了作物冠层在绿色波段和近红外波段的反射率,进而求得作物冠层的GNDVI值。用SPAD-502叶绿素计测量得到的冠层SPAD值。本研究分别对尖椒和茶树样本进行了取样,测量了其GNDVI值和冠层SPAD值并进行了拟合,得到尖椒冠层的模型为SPAD= 116.72 GNDVI-29.066其决定系数(R2)为0.8768。茶树冠层模型为SPAD=88.81GNDVI-11.32其决定系数(R2)为0.7396。两者的验证实验同样取得了良好的效果。说明该仪器的测量精度能够满足实际农业生产的需求。此外该仪器不仅适用与一种作物,而且有着适用与多种作物的潜能。(4)本研究对尖椒和茶树样本进行了混合建模,得到的模型为SPAD=104.17GNDVI-20.896其决定系数(R2)为0.7945。有良好的效果。验证实验的结果也显示误差在允许范围内。这说明该仪器有着用一个标定模型测量多种作物的潜能。(5)用VB语言为该系统编写了相应的图像处理软件,并开发了相应的仪器,能够很好的实现其功能。
[Abstract]:The excessive application of chemical fertilizer has not only caused a large amount of waste, increased the production cost of agriculture, but also caused a large number of environmental pollution and soil erosion, which is not conducive to the sustainable development of agriculture. In order to solve this problem, it is necessary to enter the idea of fine agriculture, to distinguish the crops in the field and to apply fertilizer as needed. In order to achieve this goal, it is necessary to realize the rapid nondestructive testing of crop nitrogen level in the field. Based on this requirement, a machine vision technique was developed to quickly detect the SPAD of crop canopy in the field. The specific research contents and conclusions are as follows: (1) A reference plate was made by making use of the reflectivity of crop canopy in near infrared band which is much higher than that of soil background in near infrared band. The near infrared reflectance of the reference block is located between the soil and the canopy. After graying the image of crop canopy in near infrared band, the gray value of the reference plate is used as the threshold. The crop canopy is well separated from the soil background. (2) according to the principle that the reflectivity of the object and the gray value of the pixel in the image and the intensity of the ambient light in the image are all linear positive correlation. A gradient of reflectivity is drawn on the reference board with four different colors. The linear model between reflectivity and gray value is established by using this gradient. The effect of environmental light on the calculation of crop canopy reflectivity is successfully eliminated. (3) the dynamic linear model of reflectivity and gray value is used. By calculating the average gray value of crop canopy in the image, the reflectivity of crop canopy in green band and near infrared band is obtained, and the GNDVI value of crop canopy is obtained. The SPAD values of canopy were measured by SPAD-502 chlorophyll meter. In this study, samples of prickly pepper and tea tree were sampled, their GNDVI values and canopy SPAD values were measured and fitted. The model of crown layer of prickly pepper was SPAD= 116.72 GNDVI-29.066 and the determination coefficient (R2) was 0.8768. The model of tea canopy is SPAD=88.81GNDVI-11.32 and the coefficient of determination (R2) is 0.7396. The results of the two experiments are also satisfactory. It shows that the measuring precision of this instrument can meet the demand of actual agricultural production. In addition, the instrument is not only suitable for a single crop, but also has the potential to be suitable for a variety of crops. (4) the mixed modeling of pepper and tea samples was carried out in this study. The model obtained is SPAD=104.17GNDVI-20.896 and its determination coefficient (R2) is 0.7945. Have good effect. The results of the validation experiment also show that the error is within the allowable range. This indicates that the instrument has the potential to measure many crops with a calibration model. (5) the corresponding image processing software is written for the system with VB language and the corresponding instrument is developed, which can achieve its function well.
【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：S126;TP391.41

【相似文献】