基于主动视觉的人脸检测与跟踪算法研究

系统仿真学报 ›› 2015, Vol. 27 ›› Issue (5): 973-979.

基于主动视觉的人脸检测与跟踪算法研究

董恩增, 闫胜旭, 佟吉钢

天津理工大学自动化学院复杂系统控制理论及应用重点实验室, 天津 300384

收稿日期:2014-05-05 修回日期:2014-09-15 出版日期:2015-05-08 发布日期:2020-09-01
作者简介:董恩增(1977-),男,天津,博士后,副教授,研究方向为嵌入式系统、模式识别等;闫胜旭(1985-),男,河南,硕士,研究方向为嵌入式系统、机器视觉;佟吉钢(1975-),男,天津,博士后,讲师,研究方向为嵌入式系统、智能家居等。
基金资助:
国家自然基金(61172185);天津市高等学校科技发展基金计划项目(20120829)

Research about Method of Face Detection and Tracking Based on Active Vision

Dong Enzeng, Yan Shengxu, Tong Jigang

Complex System Control Theory and Application Key Laboratory, School of Electrical Engineering, Tianjin University of Technology, Tianjin 300384, China

Received:2014-05-05 Revised:2014-09-15 Online:2015-05-08 Published:2020-09-01

摘要/Abstract

摘要： 基于融合Kalman滤波的Camshift算法与Adaboost算法相结合的方法,提出了一种基于主动视觉的人脸检测与跟踪算法。利用Adaboost算法与融合的Camshift算法实现人脸自主检测与跟踪;根据目标形心与视场中心的位置关系以及人脸区域与视场的面积关系,设计了云台控制算法。该控制算法通过对云台摄像机进行水平、垂直和变焦控制,达到自动调整云台摄像机参数的目的。基于硬件平台对所提出的主动视觉算法进行实验验证,实验结果显示,主动视觉算法具有较高的运行效率,能够实时的检测和跟踪人脸目标,扩大了摄像机的跟踪范围。

关键词: Camshift, Adaboost, 人脸检测, 主动视觉, 云台控制

Abstract: Based on the method of combining the Camshift algorithm mixing Kalman filter with Adaboost algorithm, a face-detection and tracking algorithm on the active vision was proposed. The algorithm proposed realized the face automatic detection and tracking by using Adaboost algorithm and Camshift mixed algorithm; according to the positional relation of target centroid and the view centre, and the difference of area between the region of the face target and the view, the Pan-Tilt-Zoom (PTZ) control algorithm was designed. Through the rules designed for pan, tilt and zoom control, the algorithm achieved the goal of automatic adjusting the intrinsic parameters of PTZ camera. The active vision algorithm proposed was tested in the hardware platform. The experiment results show that the algorithm has higher efficiency, and can do real-time detecting and tracking human face, which expands the range of camera tracking.

Key words: Camshift, Adaboost, target detection, active vision, Pan-Tilt-Zoom control

中图分类号:

TP391.9

董恩增, 闫胜旭, 佟吉钢. 基于主动视觉的人脸检测与跟踪算法研究[J]. 系统仿真学报, 2015, 27(5): 973-979.

Dong Enzeng, Yan Shengxu, Tong Jigang. Research about Method of Face Detection and Tracking Based on Active Vision[J]. Journal of System Simulation, 2015, 27(5): 973-979.

参考文献

[1] 夏轩, 刘华平, 许伟明, 等. 基于DSP的主动视觉系统[J]. 机器人, 2012, 34(3): 354-362.
[2] 张雪波, 路晗, 方勇纯, 等. 室外环境下PTZ摄像机全自动标定技术及其应用[J]. 机器人, 2013, 35(4): 385-393.
[3] Hyun-Cheol Choi, Unsang Park, Anil K Jain.PTZ Camera Assisted Face Acquisition, Tracking & Recognition[C]// Biometrics: Theory Applications and Systems (BTAS), 2010 Fourth IEEE International Conference on Biometrics Compendium. Washington, DC, USA: IEEE, 2010: 1-6.
[4] 石皓, 赖世铭, 刘煜, 等. 一种用于鱼眼PTZ主从监控系统的标定方法[J]. 系统仿真学报, 2013, 25(10): 2412-2417.
(Shi Hao, Lai Shiming, Liu Yu, Zhang Maojun.Calibration Method Based on Master-slave Surveillance System Composed by Fish-eye Camera and PTZ Dome Camera[J]. Journal of System Simulation (S1004-731X), 2013, 25(10): 2412-2417.)
[5] 周金广, 黄磊, 刘昌平. PTZ自主跟踪中的全景视频生成[J]. 中国图象图形学报, 2011, 16(1): 110-117.
[6] 刘士荣, 孙凯, 张波涛, 等. 基于改进Camshift算法的移动机器人运动目标跟踪[J]. 华中科技大学学报(自然科学版), 2011, 39(S2): 223-226.
[7] 相入喜, 李见为. 多特征自适应融合的粒子滤波跟踪算法[J]. 计算机辅助设计与图形学学报, 2012, 24(1): 97-103.
[8] 刘士荣, 姜晓艳. 一种改进的Camshift/Kalman运动目标跟踪算法[J]. 控制工程, 2010, 17(4): 470-474.
[9] 曹莹, 苗启广, 刘家辰, 等. AdaBoost算法研究进展与展望[J]. 自动化学报, 2013, 39(6): 745-758.
[10] 文学志, 方巍, 郑钰辉. 一种基于类Haar特征和改进AdaBoost分类器的车辆识别算法[J]. 电子学报, 2011, 39(5): 1121-1126.
[11] 朱承志. 基于OpenCV的人脸检测与跟踪[J]. 计算机工程与应用, 2012, 48(26): 157-161.
[12] Viola P, Jones M.Rapid object detection using a boosted cascade of simple features[C]// Proceedings of the 2001 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (S1063-6919). USA: IEEE, 2001, 1: 511-518.