计算机科学 ›› 2013, Vol. 40 ›› Issue (Z11): 318-319.

• 数字信息处理 • 上一篇    下一篇

基于DS-Adaboost算法的人脸检测

叶俊,张正军   

  1. 南京理工大学理学院统计与金融数学系 南京210094;南京理工大学理学院统计与金融数学系 南京210094
  • 出版日期:2018-11-16 发布日期:2018-11-16
  • 基金资助:
    本文受国家自然科学基金青年项目(11001132)资助

Face Detection Based on DS-Adaboost Algorithm

YE Jun and ZHANG Zheng-jun   

  • Online:2018-11-16 Published:2018-11-16

摘要: 针对连续Adaboost算法中平滑因子选取的不足,提出了一种动态选取平滑因子的DS-Adaboost算法,该算法对弱分类器输出中的平滑因子ε进行了动态选取,根据Wj+1Wj-1比值的大小动态地选择平滑因子,当Wj+1Wj-1>1时,εj=Wj+1,当0j+1Wj-1<1时,εj=Wj-1。实验表明,DS-Adaboost算法能较好地起到平滑的作用,使得落在同一个区间里面的正样本和负样本的比例都在可以比拟的范围内。

关键词: 连续Adaboost算法,平滑因子,权重更新,人脸检测

Abstract: Focusing on the disadvantages of the Real Adaboost algorithm selected smoothing factor.This article proposes DS-Adaboost algorithm,this algorithm dynamic select smoothing factor in weak classifier output,According to the size of the ratio of Wj+1Wj-1 dynamic selected smoothing factor.When Wj+1Wj-1>1,εj=Wj+1,0j+1Wj-1<1,εj=Wj-1.The experimental results indicate that the DS-Adaboost algorithm can better play the role of balance,make the proportion of two types of samples are within can match.

Key words: Real Adaboost algorithm,Smoothing factor,Weight update,Face detection

[1] 李月敏,陈杰,高义,等.快速人脸检测技术综述[C]∥全国16届计算机科学与技术应用学术论文集.2004:771-774
[2] Viola P,Jones M J.Robust Real-Time Face Detection[J].International Journal of Computer Vision,2001,57(2):137-154
[3] Freund Y,Schapire R E.Experiments with a New Boosting Algorithm[J].IEEE Computer Society,1996,3(4):148-156
[4] Schapire R E,Singer Y.Improved boosting algorithms usingconfidence-rated predictions[C]∥Proceedings of 11th Annual Conference on Computational Learning Theory.1998:80-91

No related articles found!
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!