欢迎访问《中国农学通报》,

中国农学通报 ›› 2022, Vol. 38 ›› Issue (34): 138-143.doi: 10.11924/j.issn.1000-6850.casb2022-0377

所属专题: 生物技术 园艺

• 农业信息·科技教育 • 上一篇    下一篇

基于图像的金桔重量预测方法研究

刘现1(), 杨航2, 李健翔2, 俞鉴签2   

  1. 1福建省农业科学院数字农业研究所,福州 350003
    2福建农林大学计算机与信息学院,福州 350003
  • 收稿日期:2022-05-08 修回日期:2022-06-15 出版日期:2022-12-05 发布日期:2022-11-25
  • 作者简介:刘现,女,1985年出生,福建福州人,助理研究员,硕士,研究方向:环境感知与智能控制。通信地址:350003 福建省福州市鼓楼区华林路188号 福建省农业科学院科技干部培训中心(数字农业研究所),Tel:0591-87869364,E-mail:fzhtlx@163.com
  • 基金资助:
    智慧农林福建省高校重点实验室开放基金项目“基于大数据的金桔智能分级模型构建及应用”(2019LSAF01);福建省基金项目“基于循环神经网络的荔枝品质分级模型优化研究”(2020J011377);福建省智慧农业科技创新团队(CXTD2021013-1);福建省农业科学院科技创新团队(CXTD2021012-3);福建省农业科学院自由探索科技创新项目(ZYTS202234);福建省数字农业科技经济融合服务平台

Multiple Weight Prediction Methods Based on Kumquat Image

LIU Xian1(), YANG Hang2, LI Jianxiang2, YU Jianqian2   

  1. 1Digital Agriculture Research Institute, Fujian Academy of Agricultural Sciences, Fuzhou 350003
    2School of Computer Science and Technology, Fujian Agriculture and Forest University, Fuzhou 350003
  • Received:2022-05-08 Revised:2022-06-15 Online:2022-12-05 Published:2022-11-25

摘要:

为提高金桔重量检测的智能化程度,基于图像进行金桔重量预测研究。在自主构建的图像采集系统试验平台上采集600张金桔图像,基于图像使用Python语言结合OpenCV计算机视觉库编程求解金桔果实大小像素值,根据实际重量数据使用最小二乘法、线性法、多项式法等进行数据拟合,构建基于金桔图像的重量预测模型。试验结果表明,基于金桔图像构建数学模型对金桔的重量进行预测具有一定地可行性和科学研究价值,其中线性法的预测模型方程为y=7.27×10-6x+23.032,多项式法的预测模型方程为y=1.199×10-15x3-1.082×10-9x2+3.053×10-4x-2.773,最小二乘法的预测模型方程为 y = 2 . 564 × 10 - 8 x ( - 9 . 688 × 10 - 6 x 2 + 2531870 . 247 )

关键词: 金桔, 图像, 重量, 预测, 模型

Abstract:

To improve the intelligent degree of kumquat weight detection, this paper studied the weight prediction based on kumquat image and proposed a new intelligent weight prediction method. We collected 600 kumquat images on the self-built image acquisition system test platform and used Python language in combination with OpenCV computer vision library to solve the size of kumquat based on the image. According to the actual weight data, we used least square method, linear method, polynomial method and other methods to fit the data, and constructed the weight prediction model based on kumquat image. The experimental results showed that it was feasible and had certain scientific research value to build a mathematical model based on kumquat image to predict the weight of kumquat. The prediction model equation of the linear method was y=7.27×10-6x+23.032, the prediction model equation of polynomial method was y=1.199×10-15x3-1.082×10-9x2+3.053×10-4x-2.773, and the prediction model equation of the least square method was y = 2 . 564 × 10 - 8 x ( - 9 . 688 × 10 - 6 x 2 + 2531870 . 247 ).

Key words: kumquat, image, weight, prediction, model

中图分类号: