摘要
在面向社区楼道或门栋监控中,人脸信息尤为重要。通过NB-IoT(窄带物联网)接入,可以满足低成本、广覆盖和多接入的需求,但是常规编码方法在窄带低码率下无法保障人脸质量。针对该问题,提出一种面向人脸业务的混合分辨率监控视频压缩方法。利用监控视频统计特性优化人脸检测速度,采用不同分辨率区别编码人脸和非人脸区域,在高倍率压缩下提高了人脸的可辨识性。实验表明,该方法相较于主流感兴趣区域编码,人脸区域PSNR提高了5.57 dB,编码速度提高了5.12倍,具备NB-IoT环境下的实用性。
In the community corridor or gate surveillance,face information is particularly important.Narrow band Internet of Things(NB-IoT)access can meet the needs of low cost,wide coverage,and multiple access.However,conventional coding methods cannot guarantee face quality in narrow band low bit rates.Aiming at this problem,a hybrid resolution surveillance video compression method for face business is proposed.We used the statistical characteristics of the surveillance video to optimize the speed of face detection,and encoded the face and non-face areas in different resolutions,which improved the recognition of face under high magnification compression.The experiments show that compared with the mainstream ROI encoding,this method improves the PSNR of face region by 5.57 dB and the coding speed by 6.12 times,and it has practical value in NB-IoT environment.
作者
肖尚武
胡瑞敏
肖晶
Xiao Shangwu;Hu Ruimin;Xiao Jing(National Engineering Research Center for Multimedia Software,School of Computer Science,Wuhan University,Wuhan 430072,Hubei,China;Suzhou Institute of Wuhan University,Suzhou 215123,Jiangsu,China)
出处
《计算机应用与软件》
北大核心
2022年第2期150-156,共7页
Computer Applications and Software
基金
公安部技术研究计划项目(2016JSYJA12)
江苏省自然科学基金项目(BK20180234)。
