计算机视觉在大米胚芽识别中的应用  

Research on Application of Computer Vision in Identifying Rice Embryo

作　　者：黄星奕[1] 吴守一[1] 方如明[1] 蔡健荣[1] 

机构地区：[1]江苏理工大学生物与环境工程学院

出　　处：《农业机械学报》2000年第1期62-65,共4页Transactions of the Chinese Society for Agricultural Machinery

基　　金：江苏省应用基础研究计划资助

摘　　要：提出了用计算机视觉系统代替人眼对大米胚芽进行自动识别的方法。通过对胚芽米的颜色特性和彩色图像的分析研究，首次提出以饱和度Ｓ作为特征参数进行胚芽和胚乳的识别，从而实现了对大米留胚率的自动检测。以建立的识别指标和方法用计算机视觉系统进行胚芽识别，结果与人工检测吻合率达８８％ 以上。留胚率的自动检测结果与人工检测结果高度吻合。The embryo ratio of rice is usually determined through visual inspection by experienced technicians. This means of measurement is imprecise, inconsistent and time consuming. A computer vision algorithm for identifying rice embryo and inspecting embryo ratio based on color analysis was developed. The saturation was extracted as color feature parameter to distinguish embryo and endosperm in rice kernel based on analyzing characteristics of rice images. By this means, embryo ratio of rice can be determined objectively and accurately. The experiments reveal that the method improves the consistence between computer vision inspection and manual inspection to over 88%. This automatic and non destructive method will soon be put into practical use in rice manufactory.

关 键 词：大米 胚芽 计算机视觉 识别 留胚率 

分 类 号：TS212.7[轻工技术与工程—粮食、油脂及植物蛋白工程] TP39.41[轻工技术与工程—食品科学与工程]