基于机器视觉的稻田灌溉用水铜离子浓度检测  

Detection of copper ion concentration in paddy field irrigation based on machine vision

作　　者：曹晓曼 臧英[1,2] 周志艳 罗锡文[1] 邢赫[1] 陈盛德 

机构地区：[1]南方农业机械与装备关键技术教育部重点实验室/华南农业大学工程学院,广东广州510642 [2]南方粮油作物协同创新中心,湖南长沙410128

出　　处：《广东农业科学》2015年第4期147-152,共6页Guangdong Agricultural Sciences

基　　金：国家"863"计划项目(2011AA100704)

摘　　要：为实时、快速地监测稻田灌溉用水的质量,确保粮食生产安全,提出了基于机器视觉的稻田灌溉用水铜离子浓度检测方法。利用搭建的机器视觉检测试验装置,设计了不同浓度铜离子溶液的机器视觉检测试验,通过提取试纸图像的特征值,运用指数回归、对数回归、2阶多项式回归和线性回归等方法建立铜离子浓度的预测模型,减少了人为认知对检测结果的影响。试验结果表明,在0～300 mg/L范围内,基于机器视觉的试纸检测水中铜的方法训练集和预测集的相关系数分别达到0.9438和0.9191,均方根误差分别为19.9563、9.7889 mg/L,误差控制在8%以下,基本上达到了对稻田灌溉用水进行实时快速检测的要求,为进一步开发实用的稻田灌溉用水监测设备提供了依据。In order to monitor the quality of paddy field irrigation water real-time and fastly, and ensure the safety of food production, a method based on machine vision for detection of copper ion concentration in the paddy field irrigation water was proposed in this paper. Detection tests based on machine vision of different concentrations of copper ion solution was designed using detection tester with machine vision. By extracting the eigenvalues of test images, methods including index regression, logistic regression, two order polynomial regression and linear regression were used to build prediction model of copper ion concentration, which reduced the influence of human cognitive on testing results. The results showed that, in the range of 0-300 mg/L, the prediction model of copper ion concentration was achieved with correlation coefficient of 0.9438 and root mean square error of 19.9563 mg/L for training set and correlation coefficient of 0.9191 and root mean square error of 9.7889 mg/L for prediction set. The relative error was kept below 8 percent, basically achieving the rapid detection of paddy field irrigation water in real time. The results of this study provide the basis for the further development of paddy field irrigation equipment utility.

关 键 词：水稻 灌溉 机器视觉 铜离子浓度 检测 

分 类 号：S274.4[农业科学—农业水土工程]