基于BP神经网络的常见中兽药中5种违禁药物显微图像识别  

Identification for microscopic images of 5 kinds of illegal medicine in common traditional Chinese veterinary medicine based on BP artificial neural network

作　　者：余晓[1] 林海丹 彭新宇[3] 翟海云[1] 崔成富 

机构地区：[1]广东药学院药科学院,广东广州510240 [2]广东省兽药饲料质量检验所,广东广州510230 [3]广东省农业科学院动物卫生研究所,广东广州510640

出　　处：《广东药学院学报》2013年第6期631-635,共5页Academic Journal of Guangdong College of Pharmacy

基　　金：广东省科技计划项目(2011B040300031)

摘　　要：目的以人工神经网络和模式识别技术为基础,对动物饲料中常添加的违禁西药进行了自动识别与分类。方法对违禁药物的显微图像进行预处理、图像分割,并提取纹理特征。根据提取的特征应用BP人工神经网络完成对5种违禁药品(呋喃妥因、呋喃唑酮、呋喃它酮、氯霉素、扑热息痛)的识别分类。结果该算法不仅能快速识别出上述5种违禁药品,且准确度比较高。结论该方法对于以上5种违禁药品能得到满意的识别结果。Objective To identify and classify illegal medicine in fodder automatically based on artificial neural network and pattern recognition techniques. Methods Preprocessing, image segmentation, feature extraction of microscopic images of medicine, and BP artificial neural network were used to complete the recognition and classification of furantoin, furazolidone, furahadone, chloramphenicol, paracetamol. Results The results show that this method can quickly identify the illegal medicine with high accuracy. Conclusion For the 5 kinds of medicine,this method could achieve a satisfactory recognition results.

关 键 词：BP神经网络 模式识别 特征提取 纹理特征 违禁药物 显微图像 

分 类 号：TP391.413[自动化与计算机技术—计算机应用技术]