Please wait a minute...
J4  2013, Vol. 47 Issue (5): 906-911    DOI: 10.3785/j.issn.1008-973X.2013.05.025
光学工程     
基于特征剪裁的AdaBoost算法及在人脸检测中的应用
孟子博, 姜虹, 陈婧, 袁波, 王立强
浙江大学 现代光学仪器国家重点实验室, 浙江 杭州 310027  
Feature pruning based AdaBoost and its application in face detection
MENG Zi-bo, JIANG Hong, CHEN Jing, YUAN Bo, WANG Li-qiang
State Key Laboratory of Modern Optical Instrumentation, Zhejiang University, Hangzhou 310027, China
 全文: PDF  HTML
摘要:

针对AdaBoost算法存在训练消耗大并且误检率较高的问题,提出一种基于AdaBoost的高效检测方法.它主要包含一种基于特征剪裁的AdaBoost算法(FPAdaBoost)和一种新的检测扫描方法——确认和跳过检测机制(CSDS).FPAdaBoost算法在每一轮训练中会根据分类误差剪裁掉一部分特征,提高算法的训练速度;而CSDS检测方法在传统的检测方法基础上引入验证和确认机制,在保证检测率的条件下有效控制误检的发生.在MIT CBCL训练集和MIT+CMU检测集上对提出的方法进行验证,结果表明,FPAdaBoost算法相比原始AdaBoost算法在性能上没有明显退化,但却大大改善了训练速度,同时CSDS检测机制的引入极大地降低误检率,提高检测结果的可靠性.

Abstract:

The AdaBoost algorithm is highly computational consuming and has high false positive rate. To deal with these problems, an efficient detection method based on AdaBoost,which consists of a Feature Pruning based AdaBoost (FPAdaBoost) algorithm and a confirmation and skipping detection scheme (CSDS), is presented in this paper. FPAdaBoost cuts off features at a certain cutting coefficient according to the classification error in each iteration of training process, which effectively speeds up the learning process and greatly reduces the computational cost. And CSDS employs verification and confirmation scheme in the conventional scanning process, which effectively eliminates false positive detections. The performance of proposed detection method was tested in face detection using the MIT-CBCL training set and the MIT+CMU test set. The results show that, compared with traditional Adaboost detection method, the training time of FPAdaBoost dramatically decreases without suffering a decline in classification capability, meanwhile the false positive rate is significantly reduced due to employing CSDS in the scanning process.

出版日期: 2013-05-01
:  TP 391.4  
基金资助:

 中央高校基本科研业务费专项资助项目(2012FZA5023);国家“十一五”科技支撑计划资助项目(2011BAI12B06).

通讯作者: 袁波(1978-),男,副教授.     E-mail: yuanbo@zju.edu.cn
作者简介: 孟子博(1986-),男,硕士生,主要从事数字图像处理、计算机视觉等方面研究.E-mail:mzbo@zju.edu.cn;
服务  
把本文推荐给朋友
加入引用管理器
E-mail Alert
RSS
作者相关文章  

引用本文:

孟子博, 姜虹, 陈婧, 袁波, 王立强. 基于特征剪裁的AdaBoost算法及在人脸检测中的应用[J]. J4, 2013, 47(5): 906-911.

MENG Zi-bo, JIANG Hong, CHEN Jing, YUAN Bo, WANG Li-qiang. Feature pruning based AdaBoost and its application in face detection. J4, 2013, 47(5): 906-911.

链接本文:

http://www.zjujournals.com/eng/CN/10.3785/j.issn.1008-973X.2013.05.025        http://www.zjujournals.com/eng/CN/Y2013/V47/I5/906

[1] FREUND Y, SCHAPIRE R E. A short introduction to boosting [J]. Journal of Japanese Society for Artificial Intelligence, 1999,14(5):771-780.
[2] SCHAPIRE R E, FREUND Y, BARTLETT P, et al. Boosting the margin: a new explanation for the effectiveness of voting methods [J]. The Annals of Statistics, 1998,26(5):1651-1686.
[3] HUANG Chang, AI Hai-zhou, LI Yuan, et al, Vector boosting for rotation invariant multi-view face detection [C]∥Proceedings of the Tenth IEEE International Conference on Computer Vision. Beijing: IEEE, 2005: 446-453.
[4]  MIDDENDORF M, KUNDAJE A, WIGGINS C, et al. Predicting genetic regulatory response using classification [J]. Bioinformatics, 2004,20(1):232-240.
[5]  OPELT A, PINZ A, FUSSENEGGER M, et al. Generic object recognition with boosting [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2006,28(3):416-431.
[6]  TORRALBA A, MURPHY K P, FREEMAN W T, Sharing features: efficient boosting procedures for multiclass object detection [C]∥ Proceedings of the 2004 IEEE Computer Society Conference on Computer Vision and Pattern Recognition. Washington: IEEE, 2004: 762-769.
[7]  VIOLA P, JONES M. Rapid object detection using a boosted cascade of simple features [C]∥ Proceedings of IEEE Computer Society of Conference on Computer Vision and Pattern Recognition. Kauai: IEEE, 2001: 511-518.
[8]  VIOLA P, JONES M. Robust real-time face detection [J]. International Journal of Computer Vision, 2004,57(2):137-154.
[9]  FRIEDMAN J, HASTIE T, TIBSHIRANI R. Additive logistic regression: a statistical view of boosting [J]. The Annals of Statistics, 2000,28(2):337-407.
[10]  贾慧星,章毓晋.基于动态权重剪裁的快速Adaboost训练算法[J].计算机学报,2009,32(2):336-341.
JIA Hui-xing, ZHANG Yu-jin. Fast AdaBoost training algorithm by dynamic weight trimming [J]. Chinese Journal of Computers, 2009,32(2):336341.
[11]  VIOLA P, JONES M. Fast and robust classification using asymmetric AdaBoost and a detector cascade [C]∥ Neural Information Processing Systems. Vancouver: NIPS, 2002: 1311-1318.
[12]  FLEURET F, GERMAN D. Stationary features and cat detection [J]. Journal of Machine Learning Research, 2008,9:2549-2577.
[13]  ROWLEY H A, BALUJA S, KANADE T. Neural network-based face detection [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1998,20(1):22-38.

[1] 徐嵩,孙秀霞,何衍. 利用直线段成像特性的摄像机畸变迭代标定方法[J]. J4, 2014, 48(3): 404-413.
[2] 谢天, 解利军, 宋广华, 郑耀. 基于平面颜色分布的增强现实自然特征注册算法[J]. J4, 2013, 47(12): 2243-2252.
[3] 杨冰, 许端清, 杨鑫, 赵磊, 唐大伟. 基于艺术风格相似性规则的绘画图像分类[J]. J4, 2013, 47(8): 1486-1492.
[4] 杨帮华, 何美燕, 刘丽, 陆文宇. 脑机接口中基于BISVM的EEG分类[J]. J4, 2013, 47(8): 1431-1436.
[5] 楼晓俊, 孙雨轩, 刘海涛. 聚类边界过采样不平衡数据分类方法[J]. J4, 2013, 47(6): 944-950.
[6] 何智翔, 丁晓青, 方驰, 文迪. 基于LBP和CCS-AdaBoost的多视角人脸检测[J]. J4, 2013, 47(4): 622-629.
[7] 刘晓芳,叶修梓,张三元,张引. 并行磁共振图像的非二次正则化保边性重建[J]. J4, 2012, 46(11): 2035-2043.
[8] 张远辉,韦巍. 在线角速度估计的乒乓球机器人视觉测量方法[J]. J4, 2012, 46(7): 1320-1326.
[9] 施锦河, 沈继忠, 王攀. 四类运动想象脑电信号特征提取与分类算法[J]. J4, 2012, 46(2): 338-344.
[10] 张大尉, 朱善安. 基于核邻域保持判别嵌入的人脸识别[J]. J4, 2011, 45(10): 1842-1847.
[11] 舒振宇, 汪国昭. 基于张量投票的快速网格分割算法[J]. J4, 2011, 45(6): 999-1005.
[12] 徐舒畅, 张三元, 张引. 基于彩色图像的皮肤色素浓度提取算法[J]. J4, 2011, 45(2): 253-258.
[13] 佘青山, 孟明, 罗志增, 马玉良. 基于多核学习的下肢肌电信号动作识别[J]. J4, 2010, 44(7): 1292-1297.
[14] 薛凌云, 段会龙, 向学勤, 范影乐. 基于FitzHughNaguno神经元随机共振机制的图像复原[J]. J4, 2010, 44(6): 1103-1107.
[15] 张远辉, 韦巍, 虞旦. 基于实时图像的乒乓机器人Kalman跟踪算法[J]. J4, 2009, 43(09): 1580-1584.