摘要
传统的人脸检测速度更多地关注检测精度,而跟踪算法多关注跟踪过程,本文综合这两种算法优点,提出改进算法实现人脸识别与跟踪。论文采用haar特征的Ada Boost算法检测面部,Eigenfaces算法实现人脸识别;视频序列中,识别的人脸作为跟踪算法的输入并实时更新;CamShift算法可以实时地跟踪人脸。实验结果显示,改进算法识别精度得到有效提高,且能达到实时跟踪的效果;人脸轮廓特征与肤色特征提高系统的鲁棒性。
The traditional face detection speed pays more attention to the detection accuracy,while the tracking algorithm pays more attention to the tracking process.This paper combines the advantages of these two algorithms and proposes an improved algorithm to achieve face recognition and tracking.The paper adopts AdaBoost algorithm of Hadar feature to detect face,and the Eigenfaces algorithm realizes face recognition.In video sequence,the identified face is used as input of tracking algorithm and updated in real time.CamShift algorithm can track face in real time.Experimental results show that the improved algorithm can effectively improve the recognition accuracy and achieve the effect of real-time tracking.The face contour feature and skin color feature can improve the robustness of the system.
作者
秦润泽
聂倩倩
Qin Runze;Nie Qianqian(Shanxi Agricultural University,Taigu Shanxi 030800,China)
出处
《山西电子技术》
2018年第3期50-53,共4页
Shanxi Electronic Technology
