节点文献
基于残差平方和的光流场正则化方法
Regularization for Optical Flow Field with Residual Sum of Squares
【摘要】 光流场的不连续性是光流场准确估计中最为重要的研究内容。提出利用最小二乘法和加权最小二乘法算法中的残差平方和来对光流场进行正则化,以减少计算窗口中含有多种模式运动的概率,较好地解决了光流场的不连续性,从而使该算法具有鲁棒性。计算结果表明,该算法具有简单实用和运算速度快的特点。
【Abstract】 Discontinuities in optical flow field is one of the most important research contents in accurately estimation of optical flow field. The regularization for the optical flow field can be carried out by use of residual sum of squares in least-square or weighted-least-square algorithm, in this way the probability of the multi-mode motion containing in the calculation window can be reduced and the discontinuities of optical flow field can be solved, thus the algorithm has robustness. The calculation results show that the algorithm has the characteristics of simplicity, practice and fast operating speed.
【Key words】 Optical flow field; Residual square error; Least square methods; Computer vision;
- 【文献出处】 光电工程 ,Opto-electronic Engineering , 编辑部邮箱 ,2001年06期
- 【分类号】TP391.4
- 【被引频次】7
- 【下载频次】176