摘要
针对酒液异物的检测问题,研究了一种基于机器视觉的智能检测方法。将该方法采集的序列图像进行加权滤波预处理;应用中值背景减除和最大熵阈值法提取出可疑异物区域;利用窗口目标搜索匹配法进行异物目标跟踪操作;根据目标跟踪路径来判断酒液中是否存在异物。实验结果表明,该方法具有较低的漏检率和误检率,较高的检测精度,能够实时有效地完成酒液异物检测。
An intelligent detection method of foreign substances in wine was studied.Firstly,the image sequences acquired by this system were pre-processed with method of weighted filtering.Secondly,the suspicious objects were extracted with methods of median background subtraction and maximum entropy segmentation.Thirdly,the objects were tracked with method of object matching in window.Finally,the product was judged to be qualified or not by the object tracking path.Experimental results demonstrate that this method has low miss detection rate and false detection rate,high detection accuracy,and it can complete real-time detection of foreign substances in wine effectively.
出处
《机械工程与自动化》
2016年第1期166-168,共3页
Mechanical Engineering & Automation
关键词
机器视觉
异物检测
序列图像
目标提取
machine vision
foreign substances detection
image sequences
object extraction
