图像识别技术在高粱不完善粒检测上的应用研究  

Application of Image Recognition Technology in Detection of Unsound Sorghum Kernels

作　　者：王军凯 赵琦 姚婷 王青松[1] 陈爽 祁坤 WANG Junkai;ZHAO Qi;YAO Ting;WANG Qingsong;CHEN Shuang;QI Kun(Yanghe Distillery Co.Ltd.,Suqian,Jiangsu 223700,China)

机构地区：[1]江苏洋河酒厂股份有限公司,江苏宿迁223700

出　　处：《酿酒科技》2025年第5期61-64,共4页Liquor-Making Science & Technology

摘　　要：不完善粒是高粱质量检测的一个重要指标,主要靠人工感官判断,主观性强、效率低。根据不完善粒检测原理,采用图像识别技术,收集不完善粒样品图像,建立不完善粒图像检测模型。结果表明,图像识别方法检测效率高,检测结果的重复性好,与国标方法无显著性差异,可以用于高粱不完善粒的日常检测。Unsound kernel is an important indicator of sorghum quality inspection,and the detection method mainly relies on manual sensory judg-ment,which is subjective and inefficient.Based on the principle of unsound kernel detection,image recognition technology is used to collect images of unsound kernel samples and establish an image detection model.The results show that the image recognition method has high detection efficiency,good repeatability,and non-significant difference compared with the national standard method.The method can be used for daily detection of unsound sorghum kernels.

关 键 词：高粱 不完善粒 图像识别 

分 类 号：TS262.3[轻工技术与工程—发酵工程] TS261.2[轻工技术与工程—食品科学与工程]