Dissertation > Excellent graduate degree dissertation topics show

Probabilistic Graphical Models for Visual Feature Analysis

Author: LiZuoXiao
Tutor: ZhangLiQing
School: Shanghai Jiaotong University
Course: Computer Software and Theory
Keywords: Probabilistic graphical models LDA Object recognition
CLC: TP391.41
Type: Master's thesis
Year: 2010
Downloads: 301
Quote: 1
Read: Download Dissertation

Abstract


The use of local visual features to solve the problem of object recognition in computer vision is a trend of the current research . The key problem is how to make good use of local visual features , so that it can faithfully reflect the image semantics . Probabilistic graphical models in the field of machine learning , on the other hand , because of its flexible modeling and sophisticated algorithms in many areas achieved a lot of success . This article focuses on the theory of probabilistic graphical model algorithm and its application in the analysis of visual features , and extend the LDA model borrowed from the field of text mining , to make it more suitable for visual problems . Modeling in computer vision from the content of this article a brief overview of focus to introduce the use of probabilistic graphical models framework model to solve visual problems . The article also reviews the theory of probabilistic graphical models and algorithms , focusing on Bayesian networks , the EM algorithm , variational inference and Gibbs sampling ; then detailed analysis the LDA model modeling , inference and learning algorithm . Then proposed in this paper the LDA one extended model - model affine invariant themes (AITM), and applied to the problem of object recognition . After simulation datasets verify the validity of AITM . The greatest contribution of this paper is to propose AITM. As an extension of the LDA , AITM abandoned in LDA \AITM not be modeled directly on the local feature position , but assumes the position of the local features of a given implicit theme by a priori position after an affine transformation obtained , and the priori position and affine modeling transform into the hidden variables . AITM the advantage with fewer parameters to characterize the spatial structure of the visual features , not easy to cause the explosion of the feature combination .

Related Dissertations

  1. Research on Intelligent Learning-Based Multi-Sensor Target Recognition and Tracking System,TP391.41
  2. Aerial Target Anti-interference Recognition and Tracking System,TN215
  3. Situaiton Assessment Using Probabilistic Graphical Models,E917
  4. Based on wavelet transform and linear subspace face recognition technology,TP391.41
  5. Based on spectral regularization of linear dimensionality reduction methods,TP391.41
  6. Video scenes based on feature extraction classification technology research,TP391.41
  7. Recommended based on social tagging systems research,TP391.3
  8. Research on the Surface Enhanced Raman Spectroscopy of Saliva for the Diagnosis of Lung Cancer,R734.2
  9. The Research and Implementation of Text Classification Based on Meta-Information and Optimization,TP391.1
  10. Intensity Distribution Investigation and Beam Shaping about the Laser Diode Array,TN248.4
  11. The Design of Classifier on Gastric Mucosa Tumor Microscopic Image,TP391.41
  12. Experimental Study on Passively Q-switched All Solid State Optical Parametric Oscillators,TN753.91
  13. Study on High Power LD Side-Pumped Nd:YAP CW Laser System,TN248
  14. Design and implementation of research and access control system based on face recognition algorithm,TP273.5
  15. Research on QoS and Congestion Control Technology over IP Multicast Networks,TP393.06
  16. The Research on Facial Expression Recognition Method,TP391.41
  17. The Study on Extractive Multidocument Summarization,TP391.1
  18. The Design of High Speed Signal Sampling and Processing System for Lidar,TN958.98
  19. LDA-based Methods for Face Recognition,TP391.41
  20. Face Recognition Method Research and Implementation,TP391.41
  21. Based on manifold learning tumor gene expression data classification,R730.4

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