基于近红外图像的嵌入式人员在岗检测系统
|
摘要
在岗检测是现代安防领域中视频分析的一个重要研究方向,应用领域非常广泛。本文设计并实现了一种嵌入式人员在岗检测系统,为了提高此嵌入式系统的运行速度,提出了改进的人脸特征点检测方法;并且为了提高系统的检测准确率,建立了一个近红外人脸样本库。该系统通过近红外摄像头采集实时图像,然后进行人脸特征点检测,获取被检测人的面部信息。根据违规行为判断准则,判断当前是否出现违规动作并且发出警报。实验结果表明:在规定条件下,系统的人脸特征点检测准确率达到了95%,针对两种异常情况的检测准确率也都超过了94%,具有良好的实时性能。
On-the-job detection is an important research direction for video analytics in the field of modern security, with a wide range of applications. This study designs and implements an embedded personnel on-the-job detection system. To improve the running speed of this embedded system, we proposed an improved face landmarks detection method, and to improve the accuracy of detection by the system, we established a near-infrared face database. This system initially collects real-time images through a near-infrared camera; subsequently, it performs face landmark detection to obtain the facial information of the detected person. According to predefined rules to identify violations, the system decides whether an illegal action has occurred and sends out alarm. The experimental results show that the accuracy of face landmark detection by the system reached 95% under the specified conditions, and the detection accuracy rate for two abnormal conditions exceeded 94%, both while maintaining a good real-time performance.
On-the-job detection is an important research direction for video analytics in the field of modern security, with a wide range of applications. This study designs and implements an embedded personnel on-the-job detection system. To improve the running speed of this embedded system, we proposed an improved face landmarks detection method, and to improve the accuracy of detection by the system, we established a near-infrared face database. This system initially collects real-time images through a near-infrared camera; subsequently, it performs face landmark detection to obtain the facial information of the detected person. According to predefined rules to identify violations, the system decides whether an illegal action has occurred and sends out alarm. The experimental results show that the accuracy of face landmark detection by the system reached 95% under the specified conditions, and the detection accuracy rate for two abnormal conditions exceeded 94%, both while maintaining a good real-time performance.
引文
[1]田国会,尹建芹,韩旭,等.一种基于关节点信息的人体行为识别新方法[J].机器人,2014,36(3):285-292.TIAN G H,YI J Q,HAN X,et al.A novel human activity recognition method using joint points information[J].Robot,2014,36(3):285-292.
[2]WANG L,XIONG Y,WANG Z,et al.Temporal segment networks:towards good practices for deep action recognition[C]//European Conference on Computer Vision,Springer,2016:20-36.
[3]WANG H,Oneata D,Verbeek J,et al.A robust and efficient video representation for action recognition[J].International Journal of Computer Vision,2016,119(3):219-238.
[4]Bay H,Tuytelaars T,Van Gool L.Surf:speeded up robust features[C]//European conference on computer vision,Springer,2006:404-417.
[5]Dlib.Dlib C++Library[DB/OL].Dlib,2018.[2018-10-25].http://blog.dlib.net/.
[6]Kazemi V,Sullivan J.One millisecond face alignment with an ensemble of regression trees[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,2014:1867-1874.
[7]山世光.Seeta Face Engine[DB/OL].Git Hub,2016[2018-10-25].https://github.com/seetaface/SeetaFace Engine.SHAN Shiguang.Seeta Face Engine[DB/OL].Git Hub,2016.[2018-10-25].https://github.com/seetaface/SeetaFace Engine.
[8]WU S,KAN M,HE Z,et al.Funnel-structured cascade for multi-view face detection with alignment-awareness[J].Neurocomputing,2017,221:138-145.
[9]Baltru?aitis T,Robinson P and Morency L P.3D constrained local model for rigid and non-rigid facial tracking[C]//Computer Vision and Pattern Recognition(CVPR),2012:2610-2617.
[10]俞红兵,乔亚.基于核DCV算法的主动近红外人脸识别方法[J].红外技术,2014,36(10):807-811.YU H B,QIAO Y.Kernel discriminative common vector method for active NIR face recognition[J].Infrared Technology,2014,36(10):807-811.
[11]恩智浦公司.i.MX6Q:i.MX 6Quad Processors-High-Performance,3DGraphics,HD Video,Arm?Cortex?-A9 Core[DB/OL].2018[2018-10-25].https://www.nxp.com/cn/products/.NXP Corp.i.MX6Q:i.MX 6Quad Processors-High-Performance,3DGraphics,HD Video,Arm?Cortex?-A9 Core[DB/OL].2018[2018-10-25].https://www.nxp.com/cn/products/.
[12]Dalal N,Triggs B.Histograms of oriented gradients for human detection[C]//Computer Vision and Pattern Recognition,2005,1:886-893.
[13]Dinges D F,Grace R.PERCLOS:A valid psychophysiological measure of alertness as assessed by psychomotor vigilance[J]//US Department of Transportation,Federal Highway Administration,Publication Number FHWA-MCRT-98-006,1998.
[14]Viola P,Jones M.Fast and robust classification using asymmetric adaboost and a detector cascade[C]//Advances in Neural Information Processing Systems,2002:1311-1318.
[15]Bradski G,Kaehler A.Learning OpenCV:Computer vision with the OpenCV library[M].Boston:O'Reilly Media,Inc.,2008.
[16]Jain V,Learned-Miller E.Fddb:a benchmark for face detection in unconstrained settings[R].Technical Report UM-CS-2010-009,University of Massachusetts,Amherst,2010.
[2]WANG L,XIONG Y,WANG Z,et al.Temporal segment networks:towards good practices for deep action recognition[C]//European Conference on Computer Vision,Springer,2016:20-36.
[3]WANG H,Oneata D,Verbeek J,et al.A robust and efficient video representation for action recognition[J].International Journal of Computer Vision,2016,119(3):219-238.
[4]Bay H,Tuytelaars T,Van Gool L.Surf:speeded up robust features[C]//European conference on computer vision,Springer,2006:404-417.
[5]Dlib.Dlib C++Library[DB/OL].Dlib,2018.[2018-10-25].http://blog.dlib.net/.
[6]Kazemi V,Sullivan J.One millisecond face alignment with an ensemble of regression trees[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,2014:1867-1874.
[7]山世光.Seeta Face Engine[DB/OL].Git Hub,2016[2018-10-25].https://github.com/seetaface/SeetaFace Engine.SHAN Shiguang.Seeta Face Engine[DB/OL].Git Hub,2016.[2018-10-25].https://github.com/seetaface/SeetaFace Engine.
[8]WU S,KAN M,HE Z,et al.Funnel-structured cascade for multi-view face detection with alignment-awareness[J].Neurocomputing,2017,221:138-145.
[9]Baltru?aitis T,Robinson P and Morency L P.3D constrained local model for rigid and non-rigid facial tracking[C]//Computer Vision and Pattern Recognition(CVPR),2012:2610-2617.
[10]俞红兵,乔亚.基于核DCV算法的主动近红外人脸识别方法[J].红外技术,2014,36(10):807-811.YU H B,QIAO Y.Kernel discriminative common vector method for active NIR face recognition[J].Infrared Technology,2014,36(10):807-811.
[11]恩智浦公司.i.MX6Q:i.MX 6Quad Processors-High-Performance,3DGraphics,HD Video,Arm?Cortex?-A9 Core[DB/OL].2018[2018-10-25].https://www.nxp.com/cn/products/.NXP Corp.i.MX6Q:i.MX 6Quad Processors-High-Performance,3DGraphics,HD Video,Arm?Cortex?-A9 Core[DB/OL].2018[2018-10-25].https://www.nxp.com/cn/products/.
[12]Dalal N,Triggs B.Histograms of oriented gradients for human detection[C]//Computer Vision and Pattern Recognition,2005,1:886-893.
[13]Dinges D F,Grace R.PERCLOS:A valid psychophysiological measure of alertness as assessed by psychomotor vigilance[J]//US Department of Transportation,Federal Highway Administration,Publication Number FHWA-MCRT-98-006,1998.
[14]Viola P,Jones M.Fast and robust classification using asymmetric adaboost and a detector cascade[C]//Advances in Neural Information Processing Systems,2002:1311-1318.
[15]Bradski G,Kaehler A.Learning OpenCV:Computer vision with the OpenCV library[M].Boston:O'Reilly Media,Inc.,2008.
[16]Jain V,Learned-Miller E.Fddb:a benchmark for face detection in unconstrained settings[R].Technical Report UM-CS-2010-009,University of Massachusetts,Amherst,2010.