一种对错误匹配点鲁棒的多单应矩阵估计方法

发布时间：2018-06-12 15:59

本文选题：多单应矩阵估计 + 近似极大似然　；参考：《机器人》2017年05期

【摘要】：针对错误匹配点干扰条件下的多单应矩阵估计问题,提出了一种对错误匹配点鲁棒的多单应矩阵估计初始化方法.该方法基于特征点对的代数误差和结构相似性约束条件,将错误匹配点剔除策略有机地融合到单应矩阵估计的过程中,在不增加计算复杂度的前提下,能够有效地剔除错误匹配点并估计出多单应矩阵的初值.结合AML-COV(approximate maximum likelihood with homography covariance)后端优化算法,本文通过仿真数据实验和真实图像实验从客观性能指标和主观视觉效果方面对算法的性能进行了验证分析.实验结果表明,本文提出的多单应矩阵估计方法能够精确、高效、鲁棒地估计出多单应矩阵的值,较好地解决了错误匹配点干扰条件下的多单应矩阵估计问题.
[Abstract]:To solve the problem of multiple homography matrix estimation under the condition of mismatch point interference, a new initialization method of multiple homography matrix estimation is proposed, which is robust to the error matching point. Based on the algebraic errors and structural similarity constraints of feature pairs, this method integrates the error-matching elimination strategy into the process of monoclinic matrix estimation, without increasing computational complexity. It can effectively eliminate the mismatch points and estimate the initial value of the multihomological matrix. Combined with AML-COV approximation maximum likelihood with homography covariance) back-end optimization algorithm, this paper verifies and analyzes the performance of the algorithm in terms of objective performance index and subjective visual effect through simulation data experiment and real image experiment. The experimental results show that the method proposed in this paper can estimate the value of multiple homography matrix accurately, efficiently and robustly, and solve the problem of multiple homography matrix estimation under the condition of mismatch point interference.
【作者单位】：火箭军工程大学控制工程系;人工智能四川省重点实验室;四川理工学院自动化与信息工程学院;
【基金】：国家自然科学基金(61203189) 陕西省自然科学基金(2015JQ6226) 人工智能四川省重点实验室开放基金(2016RYJ02)
【分类号】：TP391.41

【相似文献】