摘要
烤烟烟叶的分组依据主要有烟叶的部位和颜色,传统的分组方法主要是依靠人的感官(视觉、触觉、嗅觉)进行分组评判,这种评判方法主观随意性强,容易受到个人因素和环境因素的影响.为了克服这一缺点,利用机器视觉技术,对福建三明产区的K326烤烟,除完熟组以外的7个主组共78片烟叶提取外观特征,包括烟叶的几何特征(如长度、宽度、伸缩度、面积)和颜色特征(如色调、饱和度).将传统烟叶分组中定性的分组因素量化,再经过模糊综合评判方法判断烟叶的组别,为烟叶进一步精准化分级做准备.分组实验的正确率为91.30%,已经达到了人工分组水平.
The grouping of flue-cured tobacco leaves is mainly based on tobacco leaves of the stalk position and color.The traditional grouping heavily depends on human＇s senses such as sight,touch,and smell.This evaluation method is subjective and arbitrary,and is influenced by personal factors and environmental factors easily.In order to overcome the defects,machine vision technology is used to extract the appearance features of the SANMING FUJIAN K326 tobacco leaves,which contain 7 main groups of tobacco leaves in total 78 pieces except mellow to extract appearance features,and the geometric features,including length,width,scale degree,area and color features such as Hue,Saturation.To quantitative the groups＇ factors traditional and qualitative and judge tobacco＇s groups with the method of fuzzy comprehensive evaluation,preparation has been conducted for tobacco more precise grading.The accuracy of the experiment is 91.30%.The method has reached the level of artificial.
出处
《西南师范大学学报（自然科学版）》
CAS
北大核心
2016年第4期122-129,共8页
Journal of Southwest China Normal University(Natural Science Edition)
基金
重庆市应用开发计划项目(CSTC2014YYKFA80001)
中央高校基本科研业务费专项(XDJK2013C107)
关键词
烤烟烟叶
机器视觉
外观特征
自动分组
flue-cured tobacco
machine vision
appearance features
automatic grouping
