Issue |
JNWPU
Volume 36, Number 5, October 2018
|
|
---|---|---|
Page(s) | 942 - 948 | |
DOI | https://doi.org/10.1051/jnwpu/20183650942 | |
Published online | 17 December 2018 |
Adaptive Non-Rigid Point Set Registration Based on Variational Bayesian
基于变分贝叶斯的自适应非刚性点集匹配
1
Department of Applied Mathematics, Northwestern Polytechnical University, Xi’an
710129, China
2
The State Key Laboratory of Remote Sensing Science, Chinese Academy of Sciences, Beijing
100101, China
Received:
12
September
2017
For the existence of outliers in non-rigid point set registration, a method based on Bayesian student's t mixture model(SMM) is proposed. Under the framework of variational Bayesian, the point set registration problem is converted to maximize the variational lower bound of log-likelihood, where the transformation parameters are found through variational inference. By prior model, the constraint over spatial regularization is incorporated into the Bayesian SMM, which can adaptively be determined for different data sets. Compared with Gaussian distribution, the student's t distribution is more robust to outliers. The experimental comparative analysis of simulated points and real images verify the effectiveness of the proposed method on the non-rigid point set registration with outliers.
摘要
针对存在异常值的非刚性点集匹配问题,提出了一种基于贝叶斯混合t分布模型的匹配方法。在变分贝叶斯框架下,点集匹配问题转化为最大化对数似然的变分下界,利用变分推断确定变换参数。利用先验模型,将空间正则化约束并入贝叶斯混合t分布模型中,根据不同的点集可自适应地确定正则化参数。与高斯分布相比,t分布对异常值更加稳健。最后,在模拟点集和真实图像上的实验对比分析,验证了该方法在处理存在异常值的非刚性点集匹配问题时的有效性。
Key words: non-rigid / point set registration / variational Bayesian / student's t mixture model / outliers / robust
关键字 : 非刚性 / 点集匹配 / 变分贝叶斯 / 混合t分布 / 异常值 / 稳健
© 2018 Journal of Northwestern Polytechnical University. All rights reserved.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.