Dissertation > Excellent graduate degree dissertation topics show

Research on Key Frame Extraction Based on Clustering Algorithm and Multi-Feature Fusion

Author: WangShen
Tutor: ChengQiMin
School: Huazhong University of Science and Technology
Course: Communication and Information System
Keywords: Key Frame Extraction Multi-Feature Fusion Motion Analysis ClusteringAlgorithm
CLC: TP391.41
Type: Master's thesis
Year: 2012
Downloads: 47
Quote: 0
Read: Download Dissertation

Abstract


Key frame extraction is one of the core technologies of content-based video retrieval(CBVR). However, with regard to the characters of videos such as animation, advertisingfilms, action films, the performances of existing key frame extraction methods have limit.On one hand, the key frames extracted cannot effectively represent the original video. Onthe other hand, there exist redundant key frames. It will greatly and negatively affect thespeed and accuracy of video retrieval if using these key frames.To solve these problems, it is necessary to design a new key frame extraction methodfor such a genre of video sequence. It is also necessary to accelerate the offlinekey-frame-computing speed to make it more practical.This paper proposes a new key frame extraction method based on clustering algorithmand multi-feature fusion with regard to the type of videos of complex content, a largerange of scenarios and rich action. Firstly, similarity measure using multi-feature fusioncan describe the original video content more comprehensively. Secondly, directlyclustering the video sequence into different scenes eliminates difficulty and complexity ofshot detection and split. Thirdly, extracting the key frames based on minimum value ofmotion can represent the original video content more precisely.In addition, this paper has improved the performance of affinity propagationclustering algorithm when handling large-scale data.Finally, this paper discusses the design and implementation of thekey-frame-extraction system. Analysis and comparison of the results confirm thefeasibility of the method this paper proposed. The results show that the proposed systemhas high performance and accuracy. This research has theoretical and application value ontechnology of extraction of key frames and content-based video retrieval.

Related Dissertations

  1. Research of Video Face Recognition Based on Weighted Voting and Key-Frame Extraction,TP391.41
  2. Ultrasonic Transducer Assembly Line Planning and Its Optimization,TB552
  3. Research of Content-based Image Retrieval in Binary Trademark,TP391.41
  4. Music Driven Dance Animation,TP391.41
  5. Research of Bionic Water Strider Robot,TP242
  6. Research and Implementation of Multi-Features Fusion Based Image Retrieval Method,TP391.41
  7. Design of the Video Retrieval and Identify System of Road Diseases,TP391.41
  8. Competitive flip side swivel 360 ° motion analysis,G852
  9. Study on Multi-target Tracking Method Based on Distributed Co-evolutionary Agents,TP391.41
  10. TV sighting / tracking system design and implementation,TJ765.3
  11. Study on the Elderly Auxiliary Device Based on Kinematic Simulation Analysis,TH789
  12. Capacity Improvement of One Product Line in Company F,F273
  13. The Design and Application of Standard Operating Improvement System Based on Method Study,F273
  14. Design and Implementation of the content-based news video retrieval prototype system,TP391.41
  15. The Research on Video-based Human Upper Limb Motion Analysis Techniques,TP391.41
  16. Tracking moving objects using multi-feature fusion,TP391.41
  17. The Research of Motion Analysis and Ergonomic Evaluation on Firefighting Operation,TU998.1
  18. Research and Realization of Video Abstraction,TP391.41
  19. Video Clip Retrieval Based on the Spectral Features of Association Graph,TP391.41
  20. Video Retrieval for H.264 Compressed Video,TP391.41
  21. 3D Face Recognition Using Ifnormation Fusion,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