English

Journal of Chinese Agricultural Mechanization

Journal of Chinese Agricultural Mechanization ›› 2022, Vol. 43 ›› Issue (4): 83-89.DOI: 10.13733/j.jcam.issn.20955553.2022.04.013

Previous Articles     Next Articles

Research on soybean seedling number estimation based on UAV remote sensing technology

Li Jinyang, Zhang Wei, Kang Ye, Xu Xiuying, Qi Liqiang, Shi Wenqiang.    

  • Online:2022-04-15 Published:2022-04-24

基于无人机遥感技术的大豆苗数估算研究

李金阳1, 2,张伟1, 2,康烨1,许秀英1, 2,亓立强1, 2,石文强1, 2   

  1. 1. 黑龙江八一农垦大学工程学院,黑龙江大庆,163319;

    2. 黑龙江省保护性耕作工程技术研究中心,黑龙江大庆,163319
  • 基金资助:
    财政部和农业农村部:国家现代农业产业技术体系资助(CARS—04—PS30);北方寒地机械化保护性耕作技术创新研究团队(TDJH201808);黑龙江八一农垦大学研究生创新科研项目(YJSCX2021—Y31);大庆市指导性科技计划项目(zd—2020—64);国家大学生创业实践项目(202110223130)

Abstract: In order to improve the timeliness and accuracy of soybean seedling number estimation, a soybean seedling number estimation method based on the unmanned aerial vehicle(UAV) remote sensing technology was proposed. Unmanned aerial vehicle (UAV) was used to obtain images of soybean seedlings, different vegetation indices, and histogram equalization. The Otsu threshold algorithm was selected to extract soybean targets. Outlier removal and morphological algorithm were used to remove weed noise. The connected region analysis method was used to estimate the soybean seedling number. Combined with the field measured data, a linear regression model was established between the measured seedling number and the estimated seedling number. By comparing the extraction results of soybean targets with different methods, it is shown that the Otsu threshold algorithm combined with the overgreen index has a better effect on soybean image segmentation. The linear regression model of the measured seedling number and the estimated seedling number has a high degree of fit. The correlation coefficient R2 was 0.909 4, and the average error of emergence statistics was 0.43%. The method has a low error. Soybean seedling numbers can be quickly and accurately identified. It can provide theoretical reference for agricultural producers to obtain information on soybean seedling situations intelligently. 

Key words: soybean, unmanned aerial vehicle, image processing, seedling stages, image segmentation

摘要: 为提高大豆苗数估算的时效性和精确性,提出基于无人机遥感技术的大豆苗数估算方法。利用无人机获取大豆苗期图像,选取不同植被指数、直方图均衡化和Otsu阈值算法提取大豆目标;采用异常值剔除及形态学算法去除杂草噪声,通过连通区域分析方法估算大豆苗数。结合田间实测数据,建立实测苗数与估算苗数线性回归模型。通过对比不同方法对大豆目标提取结果表明,Otsu阈值算法结合过绿指数对大豆图像分割效果较好。实测苗数与估算苗数线性回归模型拟合度较高,相关系数R2为0.909 4,出苗统计平均误差为0.43%。该方法误差较低,能够快速准确识别大豆苗数,可为农业生产者应用智能化手段获取大豆苗情信息提供理论参考。


关键词: 大豆, 无人机, 图像处理, 苗期, 图像分割

CLC Number: