中国农业气象 ›› 2022, Vol. 43 ›› Issue (11): 935-944.doi: 10.3969/j.issn.1000-6362.2022.11.007

• 农业气象信息技术 栏目 • 上一篇    下一篇

基于机器视觉的冬小麦叶片形态测量软件开发

宫志宏,董朝阳,于红,刘布春,李春,刘涛,李军玲   

  1. 1. 中国气象局河南省农业气象保障与应用技术重点实验室,郑州 450003;2. 天津市气候中心,天津 300074;3. 天津市西青区气象局,天津 300380;4. 中国农业科学院农业环境与可持续发展研究所,北京 100081;5. 河南省气象科学研究所,郑州 450003
  • 收稿日期:2021-12-27 出版日期:2022-11-20 发布日期:2022-11-18
  • 通讯作者: 刘布春,研究员,研究方向为农业减灾和农业灾害风险管理。 E-mail:liubuchun@caas.cn
  • 作者简介:宫志宏,E-mail:gong041@126.com
  • 基金资助:
    中国气象局河南省农业气象保障与应用技术重点开放实验室开放研究基金项目(AMF201907);中国农业科学院科技创新工程(CAAS-ASTIP-2014-IEDA);河南省自然科学基金青年基金项目(202300410531)

Development of Winter Wheat Leaf Morphology Measurement Software Based On Machine Vision

GONG Zhi-hong , DONG Chao-yang, YU Hong , LIU Bu-chun, LI Chun , LIU Tao, LI Jun-ling   

  1. 1. Henan Key Laboratory of Agrometeorological Ensuring and Applied Technique, CMA,Zhengzhou 450003, China;2.Tianjin Climate Center,Tianjin 300074;3. Tianjin Xiqing District Meteorological Bureau, Tianjin 300380;4. Institute of Environment and Sustainable Development in Agriculture, CAAS, Beijing 100081;5. Henan Institute of Meteorological Sciences, Zhengzhou 450003
  • Received:2021-12-27 Online:2022-11-20 Published:2022-11-18

摘要: 随着图像处理与识别技术的快速发展,作物表型识别技术日趋成熟。为实现不同品种、不同生育期冬小麦叶片面积和面积系数的精准快速测定,依托VB.net和OpenCV在.NET平台下的图像处理封装库,研发了基于机器视觉的冬小麦叶片形态测量算法并设计开发了软件,软件可实现数字图片的畸变校准并可以同时测量多个叶片长、宽和面积。为验证软件测定效果,选取冬小麦绿色展开叶100 片,通过与人工测量的叶片长宽、WinDIAS叶面积分析系统测量的叶面积结果对比,分析图像识别方法的准确性和稳定性。结果表明,图像识别法与人工和WinDIAS测量的冬小麦叶片长、宽和面积的相关系数均≥0.975,归一化均方根误差均≤0.10%;针对数字照片畸变校准功能进行测试,对叶片水平(垂直)缩放50%且垂直(水平)斜切30°的图像校准后,其测量结果与原始图像测量结果的最大相对误差仅为2%。说明基于机器视觉的冬小麦叶片形态识别方法,可对多种畸变图像进行准确的几何校准,可作为一种可同时准确测定多个叶片面积和长宽的新方法,在农业科学测量、农情信息业务、农业气象观测业务等领域推广应用。

关键词: 冬小麦, 叶片形态, 畸变校准, RGB图像

Abstract: With the development of image processing and recognition technology, the technology of crop phenotype recognition is becoming more mature. In order to achieve accurate and rapid determination of leaf area and area coefficient of winter wheat with different varieties and different growth stages, VB.net and OpenCV software were used in the study. In the image processing and packaging library based on NET platform, a winter wheat leaf morphology measurement algorithm with machine vision was developed, and a software was designed and developed. The software can realize distortion calibration of digital images and simultaneously measure the length, width and area of multiple leaves. And then in order to verify the effect of software measurement, 100 green expanded leaves of winter wheat were selected, and the accuracy and stability of image recognition method were analyzed by comparing with the leaf length and width measured manually. The leaf area measured by WinDIAS leaf area analysis system. The results showed that the correlation coefficients of the winter wheat leaf length, width and area measured by image recognition method, manual and WinDIAS were more than 0.975 respectively, while the normalized root mean square errors were less than 0.10%. For the distortion calibration function of digital images, the maximum relative error between the measurement results and the original image is only 2% after the calibration of the image with the horizontal (vertical) scaling of 50% and the vertical (horizontal) beveling of 30°. It is suggested that the recognition method of winter wheat leaf morphology based on machine vision can accurately calibrate various distorted images, while it can be used as a new method to accurately measure the area, length and width of multiple leaves at the same time. It can be popularized and applied in agricultural scientific measurement, agricultural information service and agricultural meteorological observation service.

Key words: Winter wheat, Leaf morphology, Distortion calibration, RGB image