Spatio-Temporal based Framework for Video Retrieval System
- Paper number
IAC-17,B5,1,6,x38538
- Author
Dr. Tahir Mushtaq, University of Management and Technology (UMT), Pakistan
- Coauthor
Ms. Sana Amanat, Queen Mary College Lahore, Pakistan
- Coauthor
Mr. Ali Yousaf, Pakistan
- Coauthor
Dr. Ghulam JAFFER, University of the Punjab, Pakistan
- Coauthor
Dr. Nazish RUBAB, Institute of Space Technology (IST), Pakistan
- Year
2017
- Abstract
Due to exponential growth in multimedia data, Spatio-Temporal based various video retrieval system are designed. But these systems are fiasco to model temporal feature of videos appropriately. To cope up this limitation, a hybrid approach of various spatial feature extraction algorithms with temporal information tracking is proposed in this research paper. The hybrid/ fusion of Scale Invariant Feature Transform (SIFT), Covariant Feature Detector (CFD) and Lucas-Kanade sparse optical flow method was developed. This approach was testified and verified using standard video retrieval data set TRECVID 2014 and the results were found to be robust to noise. Moreover, the system performance was above expected value alongwith accuracy around 81\%.
- Abstract document
- Manuscript document
IAC-17,B5,1,6,x38538.pdf (🔒 authorized access only).
To get the manuscript, please contact IAF Secretariat.