Computer Engineering and Applications ›› 2009, Vol. 45 ›› Issue (7): 190-191.DOI: 10.3778/j.issn.1002-8331.2009.07.057

• 图形、图像、模式识别 • Previous Articles     Next Articles

Sub-pixel motion estimation based on two-step estimation and motion compensation

XIE Qin-lan,CAO Hui-min   

  1. College of Electronic and Information,South Central University For Nationalities,Wuhan 430074,China
  • Received:2008-09-25 Revised:2008-12-22 Online:2009-03-01 Published:2009-03-01
  • Contact: XIE Qin-lan

结合两步估算和运动补偿的子像素运动估计

谢勤岚,曹汇敏   

  1. 中南民族大学 电子信息工程学院,武汉 430074
  • 通讯作者: 谢勤岚

Abstract: For the motion estimation problems of image super-resolution,a two-step estimation method is proposed to estimate relative motion parameters between low-resolution images.Firstly,a certain method such as correlation-or distance-based method is used to estimate integer-pixels shift parameters between two images,one of them is then compensated to aligne to another.Secondly,the gradient-based method is used for the aligned images to estimate fraction-pixels shift parameters between them.Through the two estimation step,the accurate motion parameter is obtained at high speed without interpolating the low resolution images.

摘要: 针对图像超分辨中的运动估计问题,提出了一种两步估计方法,用于估计低分辨图像帧间的子像素相对运动。第一步,采用相关法或匹配法计算两幅低分辨率图像间的整像素相对位移,对其中的一幅按估计的参数进行运动补偿,第二步,对补偿后的两幅图像使用梯度法计算小数像素相对位移。通过两步计算,得到了比较精确的帧间相对运动参数。该方法不需要对低分辨率图像进行插值,以获得图像的高分辨近似,运算速度也较快。