基于标签识别的药品追溯算法设计与研究  

Design and Research of Drug Traceability Algorithm Based on Label Recognition

作　　者：霍珊[1] 刘大维[1] 徐国成[1] HUO Shan;LIU Da-wei;XU Guo-cheng(Changchun University of Chinese Medicine,Changchun Jilin 130118,China)

机构地区：[1]长春中医药大学,吉林长春130118

出　　处：《计算机仿真》2020年第10期352-356,共5页Computer Simulation

基　　金：中国图书馆学会2018年阅读推广课题(YD2018B22)。

摘　　要：药品追溯过程中标签信息易出现擦碰导致模糊,对追溯精准性造成影响,提出一种基于标签识别的药品追溯算法。在BP神经网络的基础上对药品标签图像使用滴水算法进行分割,运用K-L转换提取字符特征,将自适应调节学习率及动态调整S型激励函数相结合对药品信息精确识别;使用遗传算法按照种群特征来动态转换交叉概率和变异概率数值,利用海明距离测度和适应度距离相融合手段,保证药品信息的实时跟踪及有效查询。仿真结果表明,吞吐量达到峰值50tps,液体几乎不会对标签的读取率产生影响,药品追溯延迟时间在10s以下,可实时追踪药品流通供应链信息,有效解决药品使用安全问题。During the drug traceability, the label information is easy to be blurred, influencing the accuracy of traceability. Therefore, a drug traceability algorithm based on label identification was proposed. On the basis of BP neural network, the drop-fall algorithm was used to segment the drug label image, and K-L transformation was used to extract the features of characters. Moreover, the adaptive learning rate was combined with dynamic adjustment of S-incentive function to accurately identify drug information. According to population characteristics, the genetic algorithm was used to dynamically convert the crossover probability and mutation probability. Finally, Hemingway distance measure was integrated with fitness distance to ensure the real-time tracking and effective query of drug information. Simulation results show that the throughput reaches the peak value(50 tps), and the liquid hardly affects the reading rate of label. The delay time of drug traceability is less than 10 s, so that we can track the information of pharmaceutical distribution supply chain in real time and effectively solve the problem about safety in drug use.

关 键 词：标签识别 药品追溯 神经网络 遗传算法 滴水算法 

分 类 号：TP723[自动化与计算机技术—检测技术与自动化装置]