基于OpenCV的光缆表面喷码字符识别算法  

Characters recognition algorithm of optical cable surface inkjet code based on OpenCV

作　　者：刘梦桥 徐宁[1] LIU Mengqiao;XU Ning(School of Electronics and Optical Engineering,Microelectronics,Nanjing University of Posts and Telecommunications,Nanjing,210046,China)

机构地区：[1]南京邮电大学电子与光学工程、微电子学院,南京210046

出　　处：《光通信技术》2021年第2期6-9,共4页Optical Communication Technology

基　　金：国家自然科学基金面上项目(编号:61972209)资助。

摘　　要：人工识别光缆喷码字符弊端众多,亟需光缆自动化识别技术。针对光缆喷码点阵字符特点,提出一种在线光缆喷码字符识别系统,对模板匹配、人工神经网络和支持向量机等3种字符识别算法进行仿真研究和参数优化,并比较了这3种算法的优劣;分析了人工神经网络和支持向量机相关参数对识别准确率和训练时间的影响。仿真结果表明:相同测试集下,人工神经网络算法识别率最高,支持向量机算法在训练样本数量较少时性能出色,模板匹配算法复杂度最低。There are many disadvantages in manual identification of optical cable inkjet code characters,so the automatic identification technology of optical cable is urgently needed.Aiming at the characteristics of dot matrix characters in optical cable,an online optical cable character recognition system is proposed,the simulation research and parameter optimization of three character recognition algorithms such as template matching,artificial neural network and support vector machine are carried out,and the advantages and disadvantages of these three algorithms are compared.The influence of parameters of artificial neural network and support vector machine on recognition accuracy and training time is analyzed.The simulation results show that under the same test set,the artificial neural network algorithm has the highest recognition rate,the support vector machine algorithm has excellent performance when the number of training samples is small,and the template matching algorithm has the lowest complexity.

关 键 词：字符识别 光纤光缆 模板匹配 人工神经网络 支持向量机 

分 类 号：TN212[电子电信—物理电子学]