Dissertation > Excellent graduate degree dissertation topics show

Feature Points Matching Algorithm Based on Probabilistic Graphical Models

Author: XuJingJun
Tutor: LiuWenYu
School: Huazhong University of Science and Technology
Course: Communication and Information System
Keywords: points matching probabilistic graphical models Markov random fields non-rigid matching
CLC: TP391.41
Type: Master's thesis
Year: 2009
Downloads: 70
Quote: 0
Read: Download Dissertation

Abstract


Feature points matching play a basic and key role in computer vision and pattern recognition. It can be widely used in many areas such as image registration, medical image analysis, motion target detection and tracking, handwritten text recognition and so on. Meanwhile, feature points matching are also a difficult issue in computer vision because of the noise, which blurs the accurate relationship between two sets of points. Apart from noise, outliers and non-rigid transformation make the algorithm of points matching more and more complicated.The object of this thesis is to determine the mapping relation between two sets of points in two-dimensional space. In a great number of practical applications, there are noise and distort transformation for points matching, so rigid matching methods are inappropriate. In this case, a rotation and scale invariant feature points matching algorithm based on the probability graph model is presented in allusion to non-rigid feature points matching. The proposed algorithm can also withstand a certain degree of noise, and rigorous theoretical foundation, low complexity, good applicability are its advantages.First of all, the basic concept and key technologies of points matching are listed, the commonly used algorithms at present are compared, summarized and classified, and the advantages as well as disadvantages of them are pointed out. Then, the theoretical foundation of probabilistic graphical models, the definition of cluster and joint distribution in Markov random fields are described in detail. After converting points matching problem to map problem in probabilistic graphical models and analyzing the way to reduce the amount of calculation, a new algorithm of non-rigid points matching is proposed. At last, an improved algorithm of max subset matching is presented by using the matching threshold, which is more appropriate in practical applications.

Related Dissertations

  1. Situaiton Assessment Using Probabilistic Graphical Models,E917
  2. Approximate Variational Message Propagation for GMRF,TP18
  3. Research of Medical Image Segmentation Based on Adaptive Threshold and Markov Random Fields,TP391.41
  4. The Research and Implementation of Exemplar-Based Image Inpainting Technology Based on Markov Random Field,TP391.41
  5. Medical ultrasound image reconstruction and visualization of three-dimensional,TP391.41
  6. Markov random fields in continuous terahertz Application of Image Processing,TP391.41
  7. Omni-directional video moving target detection and tracking,TP391.41
  8. Research on Simultaneous Localization and Mapping for A Home Service Robot,TP242.6
  9. Asymmetric-Prior Author Topic Models,TP391.1
  10. The Application of Hierarchical Bayesian Space-time Model on Meteorological Data,P413
  11. Research on Feature Point Matching Algorithm of the Small Brush Robot Vision Image,TP242
  12. Several Issues in Manifold Learning,TP391.41
  13. Research of Moving Object Detection and Tracjking Based on Interest Points,TP391.41
  14. Research and Application of the Underwater Object Segmentation Algorithm Based on the Sonar Imagery,TP391.41
  15. Feature-based image registration algorithm,TP391.41
  16. Probabilistic Graphical Models for Visual Feature Analysis,TP391.41
  17. Research on Video Stabilization Technology by Feature Tracking,TP391.41
  18. Based on statistics of hyperspectral image classification technology research,TP751
  19. Research on Matching Techniques of Stereo Vision,TP391.41
  20. Research on Method of Catadioptric Panoramic Stereo Imaging and Depth Estimation,TP391.41

CLC: > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Pattern Recognition and devices > Image recognition device
© 2012 www.DissertationTopic.Net  Mobile