基于特征匹配与仿射变换的视频防抖算法  

Video Stabilization Algorithm Based on Feature Matching and Affine Transform

作　　者：庞洵[1] 李威[1] 高小伟[2] 

机构地区：[1]中国矿业大学,山东221008 [2]山东科技大学,山东221008

出　　处：《微计算机信息》2008年第12期180-182,共3页Control & Automation

摘　　要：本文提出了一种利用提取尺度不变的特征点的算法,根据视频图像的特点,建立独特的快速视频防抖的仿射变换模型,从而实现视频图像的高可靠性防抖的算法。由这种算法编写的软件可以达到消除摄像机随机抖动带来的运动噪声的效果。另外,本文还提出了用来提高处理速度的实用方案。实验表明该算法具有良好的防抖效果,能够用于视频图像的实时处理。This paper proposed an algorithm of accomplishing video stabilization reliably which based on the algorithm for extracting scale-invariant feature points and found a special model of affine transform for rapid video stabilization according to the character of video image. The soft compiled by the algorithm could achieve to an excellent effect that removing the motion noise caused by stochastic shaking of camera. Besides,this paper also put forward some practical methods to enhance processing speed. The expe...

关 键 词：视频防抖 特征匹配 仿射变换 尺度不变 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]