基于视觉的酒液中异物自动识别方法研究  

Automatic recognition method of foreign bodies in wine based on vision

作　　者：薛婷[1] 陈彦龙[1] 吴斌[2] 

机构地区：[1]天津大学电气与自动化工程学院天津市过程检测与控制重点实验室,天津300072 [2]天津大学精密测试技术及仪器国家重点实验室,天津300072

出　　处：《光电子．激光》2012年第11期2149-2153,共5页Journal of Optoelectronics·Laser

基　　金：国家自然科学基金(60902084;61172120);天津市自然科学基金(12JCQNJC02200)资助项目

摘　　要：针对白酒酒液中异物特征参数提取与自动判定方法进行了研究。根据酒液检测精度、速度要求,选取合适的图像预处理方法。针对异物的特点提出一种改进的基于轮廓跟踪行程标记算法,在标记的基础上基于单帧图像对各连通区域形态及灰度特征参数进行提取判定,剔除噪声连通区域。在连续帧中提取运动特征参数,最终判定是否为异物连通区域。实验结果表明,本文方法可以简单有效地通过多特征参数提取及融合剔除噪声区域、判别连通区域运动方向,最终识别出酒液中异物。The feature parameter extraction and judgement of foreign bodies in wine have been studied in this paper. According to the accuracy and speed requirements in wine detection, the original wine image is captured and pre-processed in the suitable method. According to the characteristics of foreign bodies in wine,an improved labelirlg algorithm, whichis based on the contour tracing and running length mark techniques is presented it1 this paper. Based on the connected component labeling, the shape and gray fea- ture parameters can be extracted and judged in a single frame of image,and part of noise areas can be re- jected. Finally,with the motion feature parameters in successive frames,foreign bodies can be judged by the motion direction of connected component. The experimental results show that the foreign bodies can be easily and effectively recognized by rejecting noise areas and judging the motion directions.

关 键 词：白酒 异物检测 连通区域标记 特征提取 识别 

分 类 号：TN247[电子电信—物理电子学]