基于自适应混合模型的药液异物视觉检测  

Foreign Substance Visual Inspection for Medicinal Solution Based on Adaptive Mixture Model

作　　者：李小迷[1] 毛建旭[1] 毛建频[2] 

机构地区：[1]湖南大学电气与信息工程学院,长沙410082 [2]抚州职业技术学院,江西抚州344000

出　　处：《计算机工程》2011年第4期212-213,223,共3页Computer Engineering

基　　金：国家自然科学基金资助项目(60835004;60872130);高等学校博士点基金资助项目(20070532048);湖南省自然科学基金资助项目(07JJ6135)

摘　　要：针对医药灌装生产线中产品异物检测的特点和要求,设计一套采用机器视觉技术的灌装药液异物自动检测系统,研究基于自适应混合模型的药液异物视觉检测方法。在对获取的药液图像预处理后,利用自适应混合高斯模型对药液序列图像数据进行检测,获得感兴趣的目标区域,采用形态学运算对目标区域进行分割。实验结果表明,该系统能准确有效地检测出药液内的异物,可满足医药灌装生产线上高速、高精度的检测要求。In this paper, according to the characteristics and requirements of the product foreign substances detection in medicine filling production line, an automatic foreign substances detection system for filling medicinal solution based on machine vision technology is designed, and the foreign substances visual inspection method based on the adaptive mixture model is studied. In the method the medicinal solution images are pre-processed, then adaptive Gaussian mixture models are applied to the original medicinal solution images to get Region-Of-Interest（ROI）. Furthermore, the mathematical morphology operator is used to target segment and foreign substances recogniton. Experimental results demonstrate that the proposed system and algorithm can detect the foreign substances in medicinal solution accurately and efficiently＋ and it can satisfy the high speed and high precision requirement in the medicine filling product line.

关 键 词：灌装药液 视觉检测 异物目标 序列图像 混合高斯模型 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]