基于贝叶斯网络的芯片质量视觉检测  

Visual inspection for tiny parts based on the Bayesian network

作　　者：丁明[1] 

机构地区：[1]徐州工程学院,徐州221000

出　　处：《现代制造工程》2013年第1期7-10,87,共5页Modern Manufacturing Engineering

基　　金：国家自然科学基金资助项目(50975276)

摘　　要：提出一种零件视觉检测方法,提取图像的暗点数、边缘点数和面积等图像特征,使用Netica软件构建具有10个属性节点和9个类节点的贝叶斯网络分类器,对各种缺陷和损坏零件进行识别,试验表明该方法具有较高的精度和效率。A kind of visual inspection method for parts was introduced. The image features such as dark pixel number,edge pixel number,area were extracted from the inmge. The Bayesian network classifier was constrncted by using the Netica software which has 10 attribute nodes and 9 category nodes. The classifier can recognize variety kinds of defective and damaged chips. Experi- mental results show that the classifier has high accuracy and efficiency.

关 键 词：损坏 缺陷 模式识别 贝叶斯网络 

分 类 号：O159[理学—数学] TP391[理学—基础数学]