摘要
通过放大视频中的微小运动,可以展现一些肉眼不可见的运动信息。原始欧拉视频放大技术(EVM方法)将输入的视频由RGB颜色空间转换为YIQ颜色空间,对视频的YIQ三个通道进行运动放大处理,这样的处理过程十分耗时。在原方法的基础上,分别提出了基于RGB颜色空间和基于Y通道的运动放大加速方法。将Y通道方法、RGB方法和EVM方法进行实验对比,结果表明:RGB方法相对EVM方法在处理速度方面有所提升,但是运动放大效果差别较大;Y通道方法相对于EVM方法在处理速度方面得到了明显的提升,而且在放大效果方面基本保持一致。
Motion magnification can be used to reveal subtle motions in video which are invisible to the naked eyes. The conventional Eulerian Video Magnification（EVM）method converts the input video signal from RGB color space to YIQ color space then magnifies the motion by processing three channels independently in the YIQ color space, but it is timeconsuming. In this paper, the RGB space based and the Y channel only based motion magnification methods are proposed separately. The experimental results of three methods（the Y channel based method, the RGB space based method and the original EVM methods）show that：in comparison with the conventional EVM method, although the processing speed of RGB based method has been improved, there are lots of defects in magnified video. However, in the Y channel based method,the processing speed is increased obviously and the magnification effect is consistent with that of the original EVM.
出处
《计算机工程与应用》
CSCD
北大核心
2015年第24期195-200,204,共7页
Computer Engineering and Applications
基金
国家自然科学基金(No.61272237)
湖北省教育厅科学技术研究计划重点项目(No.D20151204)
水电工程智能视觉监测湖北省重点实验室开放基金(No.2014KLA04)
