欢迎访问《中国农学通报》，
中国农学通报

中国农学通报 ›› 2025, Vol. 41 ›› Issue (1): 1-7.doi: 10.11924/j.issn.1000-6850.casb2023-0752

• 农学·农业基础科学 •    下一篇

冬小麦发育期识别方法研究进展

王苗苗1(), 王贝贝1(), 李明放1, 张志红2,3, 严雪2,4   

  1. 1 河南中原光电测控技术有限公司，郑州 450003
    2 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，郑州 450003
    3 河南省气象科学研究所，郑州 450003
    4 河南省气象服务中心，郑州 450003
  • 收稿日期:2023-11-08 修回日期:2024-11-01 出版日期:2025-01-05 发布日期:2025-01-01
  • 通讯作者:
    王贝贝，男，1985年出生，河南郑州人，中级职称，硕士，研究方向为系统工程与电子技术。通信地址：450003 河南郑州博学路36号 河南中原光电测控技术有限公司，E-mail：。
  • 作者简介:

    王苗苗，女，1994年出生，河南郑州人，中级职称，硕士，研究方向为计算机图像处理。通信地址：450003 河南郑州博学路36号 河南中原光电测控技术有限公司，E-mail：。

  • 基金资助:
    中国气象局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“物候气象智能化观测仪研发及其应用”(AMF202203)

Research Progress on Winter Wheat Growth Period Recognition Methods

WANG Miaomiao1(), WANG Beibei1(), LI Mingfang1, ZHANG Zhihong2,3, YAN Xue2,4   

  1. 1 Henan Zhongyuan Photoelectric Measurement and Control Technology Co., Ltd., Zhengzhou 450003
    2 Key Laboratory of Agro-meteorological Safeguard and Applied Technique in Henan Province, China Meteorological Administration, Zhengzhou 450003
    3 Henan Institute of Meteorological Science, Zhengzhou 450003
    4 Henan Meteorological Service Center, Zhengzhou 450003
  • Received:2023-11-08 Revised:2024-11-01 Published:2025-01-05 Online:2025-01-01

RichHTML

1

PDF (PC)

33

摘要：

中国作为农业大国，在农业科学技术快速发展的背景下，农业已步入高产、优质、高效的新阶段。实现作物发育期识别与观测的自动化和智能化是推动农业现代化的关键一步。文章综述现有作物发育期识别的研究现状，介绍2种冬小麦发育期自动观测识别方法，即基于归一化植被指数(NDVI)反演冬小麦发育期识别方法和基于深度学习的冬小麦发育期识别方法。以河南冬小麦为例，将2种方法发育期自动观测识别结果和人工观测结果进行对比分析，验证2种识别方法的可行性和有效性。研究发现，2种方法均有较高的识别准确度和识别效率，可提高测量效率和可靠性；在识别精度方面，2种方法在不同发育期表现各有千秋，可以互为补充。基于深度学习的发育期识别方法比植被指数反演发育期识别方法具有更好的通用性，但同时两者都要在日后对识别方法进行优化升级，以进一步提高识别精度。

关键词: 冬小麦, 发育期, 图像识别, 归一化植被指数, 深度学习

Abstract:

China is a major agricultural country. With the rapid development of agricultural science and technology, agriculture has entered a new stage of development with high yield, high quality, and high efficiency. Achieving automation and intelligent observation of crop growth period recognition is a crucial part of agricultural modernization. This paper introduced the current research status of crop growth period recognition and presented two methods for automatic observation and identification of winter wheat growth period, one based on the Normalized Difference Vegetation Index (NDVI) and the other based on deep learning. Using winter wheat in Henan as an example, the results of automatic observation and identification from both methods were compared with manual observations. The results validated the feasibility and effectiveness of both identification methods, showing high accuracy and efficiency, thereby improving measurement efficiency and reliability. In terms of identification accuracy, the two methods had their own strengths at different growth periods and could complement each other. The deep learning-based identification method demonstrated better generalizability compared to the NDVI-based method. However, both methods required optimization and upgrading in the future to further enhance identification accuracy.

Key words: winter wheat, growth period, image recognition, normalized differential vegetation index (NDVI), deep learning