摘要
随着我国经济实力的提高,大多公共场所都配备监控设备,用以对出现行人及行为的监控,通过监控识别人脸是人脸识别发展的重要方面,以后也将进一步发展。为减少视频数据的空间占比,目前市场大部分采用低分辨率的存储方式来存放视频数据,这对人脸的识别率带来了较大的影响。针对该类问题,首先利用Curvelet变换,对人脸多维空间特征进行数据采集,再利用2DPCA进行降维,最后以最近邻分类器识别,使用人脸数据图片库进行实验比对,完成人脸的识别。实验结果表明:该方法能在实时检测的基础上提高对低分辨率环境下的人脸识别效果。
With the improvement of China’s economic strength,many monitoring equipment are used to monitor predestrians in public places.The face recognition through monitoring is an important aspect of its development and will be further developed in the future.In order to reduce the spatial proportion of video data,most of the current market uses low-resolution storage to store video data,which has a great impact on the recognition rate of human faces.Aiming at the problem,this paper does as below:firstly,it uses the Curvelet transform to collect data for human faces’multi-dimensional characteristics.Secondly,it uses2DPCA for dimension reduction.Thirdly,it uses the nearest neighbor classifier to identify the face data.At last,by comparing the face database,it completes face recognition.The experimental results show that on the basis of real-time detection,the method can improve the face recognition effect under the environment of low resolution.
作者
马慧
孙万春
史君华
杨馨竹
郑集元
MA Hui;SUN Wanchun;SHI Junhua;YANG Xinzhu;ZHENG Jiyuan(Computer Basic Teaching and Research Department,Anhui Vocational College of Police Officers, Hefei 230031,China;Information Technology Center, Hefei Normal University, Hefei 230061,China;School of Computer Science and Technology, Hefei Normal University, Hefei 230061,China;College of Computer Science and Engineering,Chongqing University of Technology, Chongqing 400054,China)
出处
《重庆理工大学学报（自然科学）》
CAS
北大核心
2018年第11期162-168,共7页
Journal of Chongqing University of Technology：Natural Science
基金
2018年度安徽省高等学校自然科学研究项目(12219 zrkx2018B01)
