河南农业科学 ›› 2022, Vol. 51 ›› Issue (3): 12-19.DOI: 10.15933/j.cnki.1004-3268.2022.03.002

• 综述 • 上一篇    下一篇

基于图像处理的作物行识别算法研究进展

刁智华,闫娇楠,赵素娜,贺振东   

  1. (郑州轻工业大学电气信息工程学院,河南 郑州 450002)
  • 收稿日期:2021-08-13 出版日期:2022-03-15 发布日期:2022-05-20
  • 作者简介:刁智华(1982-),男,河南夏邑人,副教授,博士,主要从事农业机器人和精准施药方面的研究。E-mail:diaozhua@163.com
  • 基金资助:
    河南省重点研发与推广专项(202102110125);国家自然科学基金项目(62003312)

Review of Crop Row Recognition Algorithms Based on Image Processing

DIAO Zhihua,YAN Jiaonan,ZHAO Suna,HE Zhendong   

  1. (School of Electrical and Information Engineering,Zhengzhou University of Light Industry,Zhengzhou 450002,China)
  • Received:2021-08-13 Published:2022-03-15 Online:2022-05-20

摘要: 可靠的作物行识别是智慧农业机器人可靠导航的基石,可以有效减少农业机器人对农作物的损伤。基于图像处理的传统作物行识别技术包括图像预处理、特征提取、作物行拟合。综述了Hough变换法、最小二乘法、垂直投影法等传统作物行识别方法,并对其他的传统作物行识别方法进行了总结。随着智慧农业的发展,深度学习在农业领域受到越来越多的关注。更好地采集图像的各种特征,并与农业机械有效结合是深度学习作物行识别与传统方法的不同。从上述2个方面,对国内外的作物行识别算法研究做了较为系统的分析,指出基于图像处理的作物行识别算法目前存在的问题,最后根据现在的研究状况对未来的研究方向做出展望。

关键词: 作物行, 图像处理, 深度学习, 农业机器人, 智慧农业

Abstract: Reliable crop row recognition is the cornerstone of reliable navigation by intelligent agricultural robot,which can effectively reduce the damage of agricultural robot to crops.Traditional crop row recognition technology based on image processing includes image preprocessing,feature extraction and crop row fitting.The traditional crop row recognition methods such as Hough transform,least square method and vertical projection method are summarized,and other traditional crop row recognition methods are also summarized.With the development of smart agriculture,deep learning has attracted more and more attention in the agricultural field. Better collection of various features of images and effective combination with agricultural machinery are the differences between deep learning crop row recognition and traditional methods.From the above two aspects,the research on crop row recognition algorithm at home and abroad is systematically analyzed,and the existing problems of crop row recognition algorithm based on image processing are pointed out. Finally,the future research direction is prospected according to the present research situation.

Key words: Crop row, Image processing, Deep learning, Agricultural robot, Smart agriculture

中图分类号: