摘要
"摘酒"是白酒生产的重要工艺环节,现有的摘酒方法依然沿用传统的人工操作模式。摘酒工需要熟知不同酒花的特点和其对应的酒精度范围,对白酒进行分类,"看花摘酒"完全依赖摘酒工的个人经验,且容易出现人为误差等,造成产品质量不稳定。为了改善这一现状,提出采用机器视觉代替人眼的思想对不同特点的酒花进行分类。首先采用图像处理技术对原始图像进行图像滤波,hog特征提取,然后对不同类别的酒花选取适当的样本作为分类的参考,运用核支持向量机(KSVM)对数据进行分类实验,对白酒酒花的分类结果达到90%。为进一步实现自动化摘酒提供理论支持。
"Liquor picking"is an important process of liquor production.At present,the existing liquor picking methods still use the traditional manual operation mode.Wine pickers need to be familiar with the characteristics of different hops and their corresponding alcohol ranges,and classify the base wines,"looking at flower picking"completely relies on the personal experience of the pickers,and it is prone to human errors,cause product quality to be unstable.In order to solve this situation,this paper proposes to use machine vision instead of human eyes to classify hops with different characteristics.This paper first uses image processing technology to filter the original image,extract the hog feature,and then select appropriate hops for different categories.The sample is used as a reference for classification,and the kernel support vector machine(KSVM)is used to classify the data.The classification result of liquor hops reaches 90%.Provide theoretical support for the further realization of automated wine picking.
作者
潘斌
陈圩钦
姚娅川
PAN BIN;CHEN Weiqin;YAO Yachuan(School of automation and information engineering,Sichuan University of Science and engineering,Zigong 643000,China;School of physics and electronic engineering,Sichuan University of Science and engineering,Zigong 643000,China)
出处
《自动化与仪器仪表》
2021年第2期186-191,共6页
Automation & Instrumentation
基金
四川省科技厅重点项目(No.2015JY0208)
酿酒生物技术及应用四川省重点实验室项目(No.NJ2014-14)。
关键词
摘酒
图像处理
HOG特征提取
KSVM
liquor-gathering technology
image processing
hog feature extraction
support vector machine
