English

中国农机化学报

中国农机化学报

中国农机化学报 ›› 2025, Vol. 46 ›› Issue (7): 181-189.DOI: 10.13733/j.jcam.issn.2095-5553.2025.07.026

• 农业智能化研究 • 上一篇    下一篇

园区滑索智能巡检机器人设计及作物冠层温度提取方法

段小琳1，翟梦群2，张森1，韩冰1，刘宇航1，黄铝文1, 3   

  1. （1. 西北农林科技大学信息工程学院，陕西杨凌，712100； 2. 西北农林科技大学机械与电子工程学院，
    陕西杨凌，712100； 3. 农业农村部农业物联网重点实验室，陕西杨凌，712100）
  • 出版日期:2025-07-15 发布日期:2025-07-02
  • 基金资助:
    国家重点研发计划课题（2020YFD1100601）

Design of intelligent cable inspection robots and a canopy temperature extraction method in agricultural orchards

Duan Xiaolin1, Zhai Mengqun2, Zhang Sen1, Han Bing1, Liu Yuhang1, Huang Lüwen1,  3   

  1. （1. College of Information Engineering, Northwest A & F University, Yangling, 712100, China； 2. College of 
    Mechanical and Electronic Engineering, Northwest A & F University, Yangling, 712100, China； 3. Key Laboratory 
    of Agricultural Internet of Things, Ministry of Agriculture and Rural Affairs, Yangling, 712100, China）
  • Online:2025-07-15 Published:2025-07-02

PDF

摘要： 针对农业生产园区实时提取有效作物冠层温度困难的问题，以园区种植的猕猴桃与葡萄为研究对象，提出一种新型智能巡检机器人，与基于颜色空间和HSV色彩空间融合（CIELAB—HSV）的冠层温度提取方法。首先，智能巡检机器人集成数据采集设备安装在滑索上，获取作物生长环境数据、冠层红外热与可见光（TIR & V）图像。其次，采用基于轮廓角方向（CAO）并由粗到精的自动配准方法（CAO—C2F）对TIR & V图像进行配准，将双色彩空间进行融合，提取冠层区域。最后，将冠层区域图像二值化处理结果与TIR温度矩阵融合得到冠层温度。试验结果表明，改进的CIELAB—HSV方法提取冠层区域效果最优，平均精度为91.54％。同时经CAO—C2F与CIELAB—HSV方法提取的冠层温度与地面实测值相关性最高，决定系数R2为0.9525，均高于CIELAB（R2=0.9289）、HSV（R2=0.9047）以及ExG—Otsu（R2=0.8913）。且CIELAB—HSV与CAO—C2F算法结合平均耗时为3.44s，实现滑索巡检机器人在农业生产环境作业时，实时获取作物冠层温度及动态监测作物生长参数的需求。

关键词: 智能巡检机器人, 冠层温度, 图像配准, 红外热成像, 农业生产园区

Abstract: To address the challenges of effective and real-time temperature extraction of crop canopy in agricultural orchards, this study focuses on kiwifruit and grape crops grown within the orchard to propose a novel intelligent cable inspection robot and a canopy temperature extraction method based on the fusion of the CIELAB and HSV color spaces, referred to as CIELAB—HSV. First, the intelligent inspection robot is equipped with data collection devices and travels along a cableway to obtain environmental data and thermal infrared and visible light （TIR & V） images of the canopy. An automatic registration method, CAO—C2F is employed to align TIR & V images. The fusion of CIELAB and HSV dual color spaces is used to extract canopy regions. Finally, the canopy region images binary processing results are combined with the TIR temperature matrix to extract the canopy temperature. The experimental results show that the improved CIELAB—HSV method achieves the best extraction performance for canopy regions, with an average accuracy of 91.54%. Furthermore, the extracted canopy temperature, obtained through the combined application of CAO—C2F and CIELAB—HSV, exhibits the highest correlation with ground-measured values, reaching a determination coefficient R2 of 0.9525, surpassing the performance of CIELAB （R2=0.9289）, HSV （R2=0.9047）, and ExG—Otsu （R2=0.8913）. The combined CIELAB—HSV and CAO—C2F algorithms have an average processing time of 3.44 s, which enables the intelligent cable inspection robot to perform real-time canopy acquisition and dynamic monitoring of crop growth parameters in agricultural environments.

Key words: intelligent cable inspection robots; , canopy temperature, image registration, thermal infrared imaging, agricultural orchards

中图分类号: 

版权所有 © 2021《中国农机化学报 》编辑部

地址：南京市玄武区中山门外柳营100号 邮编： Tel: 025-84346270，84346296