基于机器视觉的药片表面字符检测的研究  

Research on the detection of the surface characters of tablets based on machine vision

作　　者：华鑫炎 周凯汀[1] 郑力新[2] 

机构地区：[1]华侨大学信息科学与工程学院,福建厦门361021 [2]华侨大学工学院,福建泉州362021

出　　处：《微型机与应用》2016年第7期48-50,59,共4页Microcomputer & Its Applications

基　　金：福建省科技厅项目"物联网云计算平台建设"(2013H2002);华侨大学研究生科研创新能力培育计划资助项目(1400401004)

摘　　要：针对目前药片表面字符缺陷检测难的情况,提出了一种字符缺陷检测算法。该算法根据字符的灰度值特性对Gamma映射函数进行了改进,结合γ=0.8和γ=1.2两种曲线特征,较好地分离出了药片表面字符的特征。同时,根据字符的几何特征,可以准确识别出字符存在的缺陷。实验结果表明,该算法具备较好的检测效果,检测准确率为92%。该算法在工业生产中具有一定的实用价值,可以降低药片检测的成本,提高检测效率。Aiming at the difficulty of detecting the surface characters of the tablets, a new method is proposed. According to the gray value of character, the Gamma mapping function will be improved, with the two curves ofγ =0.8 andγ = 1.2, the characteristics of the surface are better to separated. At the same time, according to the geometrical features of the characters, the defects of the characters can accurate identification. The experimental results show that the proposed algorithm has good detection effect, and the detection accuracy is 92%. The algorithm has a certain practical value in industrial production, it can reduce the cost of the detection of the tablets, and improve the detection effi- ciency.

关 键 词：机器视觉 缺陷检测 图像掩膜 GAMMA函数 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]