摘要
小麦叶锈病对我国小麦生产危害巨大,实现小麦叶锈病的监测和快速分级是进行科学生产管理的基础。针对常规图像检测技术的不足,提出一种基于红外热成像技术的快速检测和分级方法。首先,采集整株小麦样本的红外热成像图像,分别计算健康植株、潜伏期植株和显症植株的平均叶温,探明真菌入侵过程中的温度变化规律;然后,将经过直方图均衡化和中值滤波预处理的红外热成像中低于显症植株温度阈值的区域提取出来;通过温度区域划分、低温区域提取和阈值分割,计算病斑面积在整体植株热成像总面积中的百分比;最后,对病情指数进行相关分析,获得相关系数为0. 975 5,预测均方根误差为9. 79%,总识别正确率为90%。结果表明,基于红外热成像边缘检测算法的小麦叶锈病分级方法是可行的。
Wheat rust has a great harm to wheat production in worldwide. The rapid monitoring and classification of wheat rust is the basis for scientific production and management, and it is also the prerequisite to realize the treatment of wheat rust as soon as possible. In view of the shortcomings of conventional image detection algorithms, a fast detection and classification method based on infrared thermal imaging technology was proposed. Wheat samples were planted in a growth chamber at the University of Alberta, Canada. Growth chamber parameters settings were as following: temperature (max 15℃, min 11℃), photoperiod (day 12 h), light intensity (10 000 lx), RH (60%~70%). The spring wheat variety (Peace) was susceptible to rust. The infrared thermal imager brand was FLIR E6, USA. Thermal sensitivity was less than 0.06℃;FOV was less than 45°ohorizontal×34°overtical;IFOV was 5.2×10^-3 rad;IR was 160 pixels×120 pixels. The infrared thermal imaging of the whole wheat samples were collected to calculate the average leaf temperature of the healthy plants, the submersible plants and the symptomatic plants, and the temperature changes during the invasion of the fungi were detected. Infrared thermography can be used to detect leaf temperature drop caused by pathogen infection at 6d of pathogen infection incubation period, which was 7 d ahead of the naked eye observation of leaf rust spores. The Prewitt operator (PO), Sobel operator (SO), Canny operator (CO) and Laplacian operator (LO) were used to extract the edges of visible light images. The edge extraction results of PO and SO on the lesion area was not satisfactory for the complex noise processing, and the boundary gray area was seriously ghosting. LO and CO were too lean for the edges, the detection accuracy was reduced, and the background error was too large. Obviously, the direct use of conventional edge detection operators cannot meet the ultimate goal of rapid classification of diseases. Therefore, a fast detection and classification method based on infrared
作者
朱文静
陈华
李林
魏新华
毛罕平
SPANER D
ZHU Wenjing;CHEN Hua;LI Lin;WEI Xinhua;MAO Hanping;SPANER D(Key Laboratory of Modern Agricultural Equipment and Technology,Ministry of Education,Jiangsu University,Zhenjiang 212013,China;School of Agricultural Equipment Engineering,Jiangsu University,Zhenjiang 212013,China;Faculty of Agricultural,Life and Environmental Sciences,University of Alberta,Edmonton T6G 2P5,Canada)
出处
《农业机械学报》
EI
CAS
CSCD
北大核心
2019年第4期36-41,48,共7页
Transactions of the Chinese Society for Agricultural Machinery
基金
国家重点研发计划项目(2017YFD0700504)
江苏省自然科学基金项目(BK20150493)
中国博士后科学基金项目(2016M601743)
江苏大学高级人才科研启动基金项目(14JDG151)
江苏高校优势学科建设工程(苏政办发〔2014〕37号)项目
关键词
小麦叶锈病
红外热成像
边缘检测
快速分级
wheat leaf rust
infrared thermal imaging
edge detection
rapid classification
