摘要
种子精选是种子工程发展中需要解决的关键问题之一。该文设计了动态玉米籽粒品质检测分级系统,利用玉米籽粒的形态特征将种子分为4级,利用颜色特征将种子分为3级,分级合格率分别为81.8%和93.04%;设计了玉米籽粒品种识别系统,利用基于贝叶斯准则的分类器和基于支持向量机的模式识别方法,可实现农大80、农大108、高油115、农大4967、白糯6号共5个玉米品种的识别,平均识别准确率为92%;研制了玉米单倍体籽粒分拣系统,根据其颜色特征及模式识别技术进行玉米单倍体识别后使用二自由度并联机器人机构,采用气吸方式进行分拣,分拣精度为80%。
Seed inspection is one of the most urgent problems in the development of China＇s seed industry.In this paper,A dynamic quality inspection and grading system of corn kernels was developed,the morphological features were used to divide corn kernels into four grades and the color features were used to grade corn kernels into three levels,the average grading ratio of system were respectively 81.8% and 93.04%;A maize varieties recognition system based on pattern recognition was developed,the system could realize Nongda 80,Nongda 108,Gaoyou115,Nongda 4967 and Bainuo 6 with a average recognition accuracy at 92%;A haploid maize grain grading system was developed,haploid maize seeds were recognized by color features and pattern recognition technology,and a 2 degrees of freedom parallel robot in suction mode was used to sorting with a sorting accuracy of 80%.
出处
《农业工程学报》
EI
CAS
CSCD
北大核心
2010年第9期124-127,共4页
Transactions of the Chinese Society of Agricultural Engineering
关键词
品质控制
图像处理
分选
品种识别
玉米单倍体
quality control
image processing
classification
variety identification
maize haploid
