Dissertation > Excellent graduate degree dissertation topics show

Multi-feature fusion of visual tracking algorithm

Author: ZhaoLing
Tutor: FengZuo
School: Huazhong University of Science and Technology
Course: Communication and Information System
Keywords: Visual Tracking Block-Based Tracking SVM classifier based tracking Gaussian mixture model Multiple Features
CLC: TP391.41
Type: Master's thesis
Year: 2011
Downloads: 52
Quote: 0
Read: Download Dissertation

Abstract


With the computer and network technology in the world within the scope of universal application , a large number of video surveillance resources have emerged , and how efficient use of these resources has become a priority. Visual tracking technology is the basis of related applications , the field of computer vision has become the most important direction of development. This paper introduces the visual tracking application , discusses the focus of visual tracking , pointing out the research background , purpose and content. Tracking algorithm then introduced the classification , as well as representatives of the classic algorithm introduces belong Bayesian filter Kalman filter and particle filter, as well as another classic mean shift algorithm . More concern in recent years while a new tracking algorithm based on block -based classifier tracking algorithm, based on level set tracking algorithm as well as a combination of multi- tracker tracking algorithms have also been introduced . On this basis , and pointed out that the current algorithm analyzes some existing problems and shortcomings , presented our research directions and contributions. This paper presents the appearance of the model based on adaptive block tracking algorithm , the introduction of Gaussian mixture model to model the target and template updates. The algorithm can be changed according to the appearance of the moving object , position and occlusion phenomena adaptive template to implement robust tracking. Through a comprehensive qualitative and quantitative experimental results demonstrate that the proposed algorithm is superior to existing tracking effect similar algorithms . This paper also proposed a multi- feature fusion classifier based tracking algorithm, which will be tracked objects as positive examples , the background and other non- target areas as counter-example , if we can use the classifier ( eg SVM) to two class effectively separate, robust tracking can be achieved . The effect depends largely on the characteristics of the object used to express whether an object can reflect the uniqueness and difference. This paper presents a method using a combination of structure, texture and color features to express the tracking target, where structural component with edge direction histogram , texture component based on texture primitive way of expressing the coefficient vector . Experimental results show that the proposed algorithm is better than existing similar tracking algorithms. Finally, the main research and work have been summarized , and the need to further study and work.

Related Dissertations

  1. Detection and Tracking of Moving Object in Complex Background,TP391.41
  2. Study on Visual Target Detection Based on SVM,TP391.41
  3. Research of Moving Object Detection and Tracking Technology,TP391.41
  4. Video surveillance system moving target detection algorithm,TP391.41
  5. Research of Hybrid EM Algorithm Based on Gaussian Mixture Models and Its Clustering Application,TN911.7
  6. Design and Research for a Kind of Hemispheric Differedtial Motion Spherical Robot,TP242
  7. The Research of Unstructured Road Detection Algorithm Based on Color Features and B-spline and Its Application,TP391.41
  8. Wideband audio technology nonlinear band broadening,TN912.3
  9. Content-based image retrieval key technology research,TP391.41
  10. Research on the Key Technologies of Intelligent Transportation Video Surveillance System,TP391.41
  11. Distributed sound source localization and tracking algorithm,TN912.3
  12. Embedded target detection and tracking system design and algorithm implementation,TP391.41
  13. Image Spam Detecting Based on Combinatorial and Statistical Classifier,TP391.41
  14. Key Algorithm in High Quality Voice Conversion System,TN912.3
  15. Research on Speaker Recognition and Its Robustness in Noise Environment,TN912.34
  16. Detection instead of classification of brain-computer interface research,TN911.7
  17. Conference room environment based on audio and video information fusion speaker recognition,TN912.34
  18. Research on Language IDE NT Ification,TN912.34
  19. GMM - UBM model - based language recognition,TN912.34
  20. Research on Moving Object Detection Technology for Intelligent Visual Surveillance,TP391.41
  21. Research of Feature Extraction for Voice Based on Wavelet Transform,TN912.34

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