Home About us Editorial board Search Ahead of print Current issue Archives Submit article Instructions Subscribe Contacts Login 
  • Users Online: 44
  • Home
  • Print this page
  • Email this page
Year : 2019  |  Volume : 5  |  Issue : 2  |  Page : 76-84

Robust point set registration method based on global structure and local constraints

College of Electronic and Information Engineering, Tongji University, Shanghai, China

Correspondence Address:
Yufei Chen
College of Electronic and Information Engineering, Tongji University, No. 4800, Cao'an Highway, Shanghai 201804
Login to access the Email id

Source of Support: None, Conflict of Interest: None

DOI: 10.4103/digm.digm_10_19

Rights and Permissions

Background and Objectives: Point set registration is a very fundamental problem in computer vision. The registration problem can be divided into rigid and nonrigid registration. The transformation function modeling of rigid registration is simpler, whereas the nonrigid registration is better to solve the practical problems. Materials and Methods: We proposed a robust point set registration method using both global and local structures. Here, we use a popular probability model, Gaussian mixture model, to preserve the global structure of point set. Then, we designed a local constraint provided by some neighboring points to maintain the local structure of the point set. Finally, expectation–maximization algorithm is used to update model parameters in our method. Results: First of all, we carried out experiments on the synthesized data, which included four degradation cases: deformation, noise, outlier, and rotation. By comparing the mean and standard deviation of registration errors with the several state-of-the-art methods, our method was proved to have stronger robustness. Then, we conducted experiments on real retinal fundus images, aiming to establish reliable feature point correspondence between the two images. The experimental results show that we perform better when the two images have larger shooting angles and more noises. Conclusions: The Gaussian mixture protects the global structure of the point set, and the local constraints make full use of the local structure, which makes our method more robust. Experiments on synthetic data prove that our method obtains superior results to those of the state-of-the-art methods. Experiments on retinal image data show that our method also performs very well in practical applications.

Print this article     Email this article
 Next article
 Previous article
 Table of Contents

 Similar in PUBMED
   Search Pubmed for
   Search in Google Scholar for
 Related articles
 Citation Manager
 Access Statistics
 Reader Comments
 Email Alert *
 Add to My List *
 * Requires registration (Free)

 Article Access Statistics
    PDF Downloaded130    
    Comments [Add]    

Recommend this journal