粮油食品科技

中国科技核心期刊

基于卷积神经网络算法处理的稻米参数评定分析仪
DOI:
作者:
作者单位:

作者简介:

通讯作者:

基金项目:


A Morphological Analyzer for Rice Shape Based on CNN Method
Author:
Affiliation:

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
    摘要:

    基于电荷耦合元件(Charge-coupled Device,CCD)摄像头的嵌入式硬件平台和卷积神经网络(Convolutional Neural Networks,CNN)算法原理,研发对采集图像进行灰度和卷积处理的稻米参数评定分析仪,通过CNN算法分析出米粒边缘及透明度比较高的部分,并根据预设的第一阈值进行边缘切割,分离出单个米粒,并计算该米粒的总像素数、最长直线像素数、最宽直线像素数,而后计算图像中小于第二设定阈值的像素数,用该像素数除以总像素数计算垩白度,将计算出来的长度、宽度、长宽比、垩白度与国家标准比对,给出所测试的稻米参数,通过对8组实验样品测试分析结果表明,单次样品绝对误差值为0.02,相对误差值为–0.31%,相关数据可上传到上位机用于品质管控,本稻米参数评定分析仪原理和算法具有一定创新性,能缩短检测时间,提高检测精准度,具有一定的应用价值。

    Abstract:

    A morphological analyzer for rice shape with gray and convolution processing of collected images was developed in this article based on embedded hardware platform of charge coupled device (CCD) camera and algorithm principle of convolutional neural networks (CNN). The camera gets the morphological image of rice, and translates it to gray level image, then the analyzer processes the image data using CNN network by the first and second threshold, the analyzer can get the edge of a race in the image, and gains all the pixels of the rice, counts the length and wide of the rice, gets the square of the race image. Then according the state standard, it gives the grade of the testing race. Eight groups of experimental samples were tested and analyzed, and the results showed that the absolute error of a single sample was 0.02, and the relative error was –0.31%. Meantime, the analyzer can uploaded the relative data to computer for quality control. The principle and algorithm of the rice parameter evaluation analyzer in this research is innovative and has a certain application value, which can shorten the detection time and improve the detection accuracy.

    参考文献
    相似文献
    引证文献

当期目录


年第卷第

文章目录

过刊浏览

引用本文

步东伟.基于卷积神经网络算法处理的稻米参数评定分析仪[J].粮油食品科技,2021,29(4):187-191.

复制
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:
  • 最后修改日期:
  • 录用日期:
  • 在线发布日期: 2021-07-29
  • 出版日期: