摘要
一维投影模型运动参数估计是图像序列运动分析的关键技术,本文针对经典最小平方估计算法计算量大的缺点,提出了一种改进的一维投影模型运动参数估计算法。该方法通过对非线性最小平方误差函数的线性化,重新定义了目标函数,采用Newton最优化算法求解参数矢量,得到只与已知特征点对应关系有关的海森矩阵后,利用线性算法求解图像序列相邻两帧间一维投影模型的运动参数。实验结果表明,该算法在保证精确度的条件下,提高了计算效率。
The motion parameters estimation of one-dimensional projective model is a key technique in im- age sequence analysis. Aiming at the shortcoming of the algorithm for Classical Least-square estimation, that is, huge computational cost, an improved algorithm for motion parameters estimation of one-dimen- sional projective model is proposed in this paper. Firstly, the algorithm redefines the objective function by lineariziug the non-linear least-square error function. Then, the Newton optimization-algorithm is used to get the vector of the parameter. As a result, the Hessian is only related to the feature-correspondences that we have known. At last, the motion parameters of one-dimensional projective model are linearly solved. Experimental results show that the proposed algorithm ensure the precision and improve the computational efficiency.
出处
《中国体视学与图像分析》
2009年第2期178-181,共4页
Chinese Journal of Stereology and Image Analysis
基金
国家自然科学基金(60872144)
关键词
运动参数估计
一维投影模型
最小平方误差
Motion parameter estimation
One-dimensional projective model
Least-square error