Non-Rigid Point Set Registration via Adaptive Weighted Objective Function

作     者：Yang, Changcai Liu, Yizhang Jiang, Xingyu Zhang, Zejun Wei, Lifang Lai, Taotao Chen, Riqing 

作者机构：Fujian Agr & Forestry Univ Coll Comp & Informat Sci Digital Fujian Res Inst Big Data Agr & Forestry Fuzhou 350002 Fujian Peoples R China Fujian Agr & Forestry Univ Fujian Prov Univ Key Lab Smart Agr & Forestry Fuzhou 350002 Fujian Peoples R China Wuhan Univ Elect Informat Sch Wuhan 430072 Hubei Peoples R China 

出 版 物：《IEEE ACCESS》 (IEEE Access)

年 卷 期：2018年第6卷

页      面：75947-75960页

核心收录：

基　　金：National Natural Science Foundation of China [61501120, 61701117, 61702101] Outstanding Youth Research Talent Cultivation Program Colleges and Universities in Fujian Province Fujian Agriculture and Forestry University [XJQ201514] Natural Science Fund of Fujian Province [2017J01736] 

主　　题：Registration expectation-maximization (EM) algorithm thin-plate spline (TPS) objective function mixture model point set 

摘      要：Non-rigid point set registration is a fundamental problem in many fields related to computer vision, medical image processing, and pattern recognition. In this paper, we develop a new point set registration method by using an adaptive weighted objective function, which formulates the alignment of two point sets as a mixture model estimation problem. The correspondences and the transformation are jointly recovered by using the expectation-maximization algorithm to obtain the promising results. First, the correspondences are established using local feature descriptors, and the adaptation parameters for the mixture model are computed from these correspondences. Then, the underlying transformation is recovered by minimizing the adaptive weighted objective function deduced from the mixture model. We demonstrate the advantages of the proposed method on various types of synthetic and real data and compare the results against those obtained using the state-of-the-art methods. The experimental results show that the proposed method is robust and outperforms the other registration approaches.