Dissertation > Excellent graduate degree dissertation topics show
Research of Point Set Pattern Matching Based on Spectral Graph Method
Author: WangSongTao
Tutor: PanZuo
School: Wuhan University of Technology
Course: Computer Science and Technology
Keywords: Conversion and the corresponding Point pattern matching Spectral graph theory Neighboring matrix Laplacian matrix Image texture
CLC: TP391.41
Type: Master's thesis
Year: 2011
Downloads: 85
Quote: 0
Read: Download Dissertation
Abstract
|
Point pattern matching computer vision and pattern recognition based on important issues , is currently a hot topic of concern in various fields . It is a widespread problem , not limited to the field of computer vision , Air -bit and posture estimation , target recognition , remote sensing image registration , medical image registration calculation has a wide range of biological and chemical applications . The spectral graph theory is applied to the study of the subject has high computational efficiency , better advantage . By Scott and Longuet-Higgins, Shapiro and Brady both point mode matching the lack of analysis of the spectrum , a new point pattern matching spectrum analysis method . A core part of the method is to construct a new neighboring matrix , the better the results obtained in the case of this method in more substantial image under the affine transformation and the point of jitter in the previous two methods , on the complexity of the algorithm with Spectrum diagram method has the advantage of efficiency than the iterative methods proposed in recent years , some high . The new methods have been more comprehensive testing , including tests on synthetic and real data , we have achieved good results . In this paper, based on previous literature spectral graph theory matching algorithm , make some improvements and refining the framework of an algorithm , the algorithm is divided into two steps , the first step is to choose the combination of distance function , the framework itself provides a group improved distance function , also depending on the application needs additional function . The final product of the distance function for the selected function , this product can be highlighted the role of the distance function for specific matching demand . The second step is to build distance function Gaussian weighted adjacent matrix or Laplacian matrix for the SVD decomposition , matching relation . The advantage of this framework can be generated in accordance with the needs of the specific application of a variety of algorithms have better flexibility and adaptability , implementation , every step of the generated code can be reused , easy programming .
|
Related Dissertations
- On the Ordering of Trees by the Laplacian Coefficients,O157.5
- Decomposition Design of Decentralized H_∞ Controllor for a Class of Multiagents System,TP13
- Research on Infrared Multi-camera Optical Tracking System for Surgical Robot,TP242
- Research on Point Pattern Matching Algorithm Based on Invariant Sequence,TP391.41
- Laplacian Spetral Radius and Signless Laplacian Spetral Radius of Graphs,O157.5
- Synchronization and Passivity Analysis of Complex Delayed Dynamical Networks,O157.5
- Mono-vision Based Aerocraft Pose Measurement Technology,TP391.41
- The Minimal Signless Laplacian Spectral Radius of Graphs with Grven Diameter,O157.5
- The Critical Group of Some Kinds of Corona,O157.5
- The Automatic Road Extraction Technology Based on Fuzzy Connectedness,TP751
- On Graphs Determined by Their Spectrum and Angle,O157.5
- On the Sandpile Group of Two Kinds of Graphs,O157.5
- The spectral radius of connected graphs and diagrams Las pull spectral estimation spectral radius,O157.5
- Based on micro-features fingerprint recognition algorithm,TP391.41
- The Molecular Dock of Combination Pharmacophore’s Assessment Method and Their Potential Use in the Development of New Drugs,TP391.9
- The Algebraic Connectivity of One Type Weighted Graph,O157.5
- Research on Algorithms of Point Pattern Matching Based on Graph Spectral Theory,TP391.41
- Largest and the second small figure Laplace eigenvalues,O157.5
- Research on the Image Matching Algorithm of LED Chip Check and Sorting,TP391.41
- Techniques on Interest Points Detection and Image Matching,TP391.4
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
|