摘要
介绍了数字图像稳像算法的整个流程,包括三部分:运动估计、运动决定和运动补偿。重点介绍了运动估计算法和运动补偿中的滤波算法;详细阐述了几种最基本的运动估计算法,其它的运动估计算法可以通过这些基本算法延伸发展得到;列出了三种常用的滤波算法:中值滤波、kalman滤波和二次B样条曲线法,应用于不同的场合;提出了两种方法:裁减和拼接,用来解决当前帧补偿后出现的空白区域问题。最后,展示了一种基于投影估计和kalman滤波的稳像算法的试验结果。通过这些方法的介绍,可以对稳像算法有更直观的了解,探讨了进一步研究的重点。
The process of digital image stabilization,which includes motion estimation,determination,and motion compensation,is introduced.Filter methods in the motion compensation are mephatically introduced.Several basic motion estimation algorithms are circumstantiated,from which other motion estimation algorithm could be developed.Three cmmonly used filter algorithms i.e.the median filter,the kalman filter and the two-order B-spline-based filter are listed.Trimming and mosaic,the ways for solving the problem of filling in the blank field after compensation are introduced.Also,the experiment results of projection estimation ＆ kalman filter based stabilization algorithm are showed in this paper.Further challenges are foreseen from these introductions.
出处
《计算机仿真》
CSCD
2008年第7期200-204,共5页
Computer Simulation
关键词
电子稳像
运动估计
运动补偿
运动模型
Electronic image stabilization
Motion estimation
Motion compensation
Motion model
