摘要
针对室内灯光环境下,视频人脸检测存在的偏色及复杂背景干扰问题,提出了一种结合视频帧预处理的人脸检测方法。首先采用灰度世界算法校正视频帧偏色,然后在YCbCr色彩空间的Cr通道内,应用OTSU二值化算法实现视频帧肤色分割,以降低复杂背景对人脸检测算法的干扰,最后使用AdaBoost算法对分割后视频帧内容进行人脸检测。实验结果表明,该方法在三类人脸检测评价指标及运行速度方面更优,可在室内灯光环境中作为一种视频人脸自动检测技术使用。
Aiming at the problem of color deviation and complex background interference in video face detection under the indoor lighting environment,a video frame preprocessing method is proposed.The gray world algorithm is first used to correct the color cast of the video frame.In the Cr channel of the YCbCr color space,the OTSU binarization algorithm is applied to realize the video frame’s skin color segmentation to reduce the interference of the complex background on the face detection algorithm.The AdaBoost algorithm is used to detect the face of the segmented video frame contents.The experimental results show that the proposed method is better in the evaluation index and running speed of three types of face detection,and it can be used as a video face automatic detection technology in the indoor lighting environment.
作者
孙鹏
解梦达
郎宇博
魏丹妮
单大国
SUN Peng;XIE Meng-Da;LANG Yu-Bo;WEI Dan-Ning;SHAN Da-Guo(School of Information Technology and Intelligence,Criminal Investigation Police University of China,Shenyang Liaoning 110035;Video Investigation Unit,Yuhong Department,Shenyang Public Security Bureau Shenyang,Liaoning 110035;National Laboratory of Sources of Physical Evidence on the Crime Scene,Beijing 110000)
出处
《中国刑警学院学报》
2021年第3期115-122,共8页
Journal of Criminal Investigation Police University of China
基金
2017年度国家重点研发计划专项(编号:2017YFC0822204)
2017年度现场物证溯源技术国家工程实验室开放课题(编号:2017NELKFKT09)
2020年度中国刑事警察学院研究生创新能力提升项目(编号:2020YCZD03)。
