首页 | 本学科首页   官方微博 | 高级检索  
     检索      

功率管红外热图像统计识别与分析研究
引用本文:万寿红,李新华.功率管红外热图像统计识别与分析研究[J].安徽大学学报(自然科学版),2006,30(1):15-19.
作者姓名:万寿红  李新华
作者单位:1. 中国科学技术大学,计算机科学技术系,安徽,合肥,230027
2. 安徽大学,计算智能与信号处理教育部重点实验室,安徽,合肥,230039
摘    要:通过对半导体功率管管壳的红外热图像进行自动聚焦,提取对不同焊接情况的图像有明显分类效果的特征参数,对训练样本进行统计分析,形成专家知识库,运用最小特征加权街区距离分类器,快速、准确地实现自动识别分类.对焊接有缺陷的情况,利用基于AI的自动阈值选择方法,分割出缺陷区域,并对缺陷做定量分析.

关 键 词:红外热图像  模式识别  AI  阈值选取
文章编号:1000-2162(2006)01-0015-04
收稿时间:2005-05-13
修稿时间:2005-05-13

Statistical recognition and analysis research in the infrared thermal images of power package
WAN Shou-Hong,LI Xin-Hua.Statistical recognition and analysis research in the infrared thermal images of power package[J].Journal of Anhui University(Natural Sciences),2006,30(1):15-19.
Authors:WAN Shou-Hong  LI Xin-Hua
Abstract:Expert knowledge database can be created by statistic analysis of training sample,after auto-matic focusing and extracting feature parameters which are very useful to classify different solder condition in infrared thermal images of the semiconductor power package.In this paper,a minimum feature-weighted block distance classifier is used to realize recognition quickly and accurately.Quantitative analysis of flaws is realized by using the method of selecting threshold automatically based on AI to segment the flows for the soldering flows condition.
Keywords:infrared thermal image  pattern recognition  artificial intelligence  threshold selection
本文献已被 CNKI 维普 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号