摘 要：
在育种过程中,挑选一致性好的种子是一项繁琐而艰苦的工作,用机器视觉代替人的视觉可以克服人眼疲劳等因素对种子进行筛选所造成的误差。为此,采用OpenCV图像处理软件设计种子识别软件,对种子的面积、周长、圆形度、直径、延伸度等特征参数进行提取。根据与样本参数的对比决定是否合格,并通过串口送至下位机,控制下位机完成种子的分拣工作。该系统设计了自学习功能,样本种子可自由选择,并可自由控制筛选误差范围。经试验验证,该系统能够准确地判断玉米和小麦种子是否合格,判断速度为60粒/min。
译 名：
Software Design of Automatic Seed Identify System
作 者：
Ji Zhongyu,Gong Linong,Huang Tingjian(Mechanical and Electrical Engineering College,Qingdao Agriculture University,Qingdao 266109,China)
关键词：
seed selection;image processing;OpenCV
摘 要：
In breeding experiments,selection of seeds in the same shape is a tedious and laborious work.Using machine vision to replace human vision can overcome seeds selection errors caused by the human eye fatigue and other factors.This paper uses the OpenCV image processing software to design a seed selection system.This system can calculate seeds' characteristics include area,perimeter,diameter,roundness and elongation.Comparing with the sample seed's characteristic,seed selection system could determine the seeds whether qualified.The determination result will be sent to the lower pc through the serial port,which will control lower pc to complete sorting seeds work.The system designs of self learning function.Sample seed can be chose freely and the accuracy can be controlled freely.It is proved by the experiment that the system can select the maize seeds and wheat seed accurately.Judge speed is about 60 grains / min.
