|Title:||Video based location search using SIFT points|
|Subject:||Hong Kong Polytechnic University -- Dissertations|
Digital video -- Data processing
Information storage and retrieval systems -- Data processing
|Department:||Department of Computing|
|Pages:||x, 84 leaves : ill. ; 31 cm.|
|Abstract:||With the exponential growth of web-based application and technology, the development of multimedia transmission and information retrieval methods grows fast. Video search website also becomes increasingly popular. However, the information retrieval methods used in video search still depend on the context of the attached description rather than the video content itself. This dissertation presents a brand-new method for content-based video search method using SIFT descriptors as the visual primitives to describe the visual property of video and perform matching operation to the other videos or images. The key technical challenges are the robustness of the SIFT based object matching in video sequences, and the computational complexity associated with the large scale of the repository. This dissertation paper solved this problem by develop a spatio-temporal pruning method for reducing the volume of SIFT points in the video repository. Besides that, the SIFT repository also deploy an efficient indexing method combined with PCA projection algorithm and indexing method to quantize the SIFT points. Simulations demonstrated the robustness of the PCAed and indexed SIFT points in building identification and overall effectiveness of the proposed solution on a small scale trial. The experiment test 3 set of different query format, the small resolution video query is proved to have the best recognition performance. Tagging and integration with large scale video repositories like YouTube, Tudou are underway with more interesting applications.|
|Rights:||All rights reserved|
Files in This Item:
|b23910860.pdf||For All Users (off-campus access for PolyU Staff & Students only)||2.38 MB||Adobe PDF||View/Open|
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item: