计算机工程与应用 ›› 2009, Vol. 45 ›› Issue (31): 165-167.DOI: 10.3778/j.issn.1002-8331.2009.31.049

• 图形、图像、模式识别 • 上一篇    下一篇

基于Cascade的增量式组合分类算法研究

欧吉顺1,朱玉全1,陈 耿2,刘 晟1   

  1. 1.江苏大学 计算机科学与通信工程学院,江苏 镇江 212013
    2.南京审计学院 省级审计信息工程重点实验室,南京 210029
  • 收稿日期:2008-06-18 修回日期:2008-10-23 出版日期:2009-11-01 发布日期:2009-11-01
  • 通讯作者: 欧吉顺

Research on combined classifier algorithm based on cascade

OU Ji-shun1,ZHU Yu-quan1,CHEN Geng2,LIU Sheng1   

  1. 1.School of Computer Science and Telecommunications Engineering,Jiangsu University,Zhenjiang,Jiangsu 212013,China
    2.Jiangsu Key Laboratory of Audit Information Engineering,Nanjing Audit University,Nanjing 210029,China
  • Received:2008-06-18 Revised:2008-10-23 Online:2009-11-01 Published:2009-11-01
  • Contact: OU Ji-shun

摘要: 利用Learn++思想对Cascade组合分类器进行了改进,提出了一种基于Cascade的增量式组合分类算法,并将之应用到肝脏图像的分类中。实验结果表明,与原有组合分类器相比,该增量式组合分类方法可以在保证分类准确度的前提下有效地提高新增样本的学习效率。

关键词: 多分类器组合, 增量式更新, Learn++, Boosting

Abstract: The Learn++ method is applied to improve the Cascade combined classifier,and is applied to lung images classification.The experiment results show that the incremental combined classification method can obviously improve the efficiency at the precondition that assure the accuracy compared with combined classifier.

Key words: multiple classifiers combination, incremental updating, Learn++, Boosting

中图分类号: