Video Searching and Retrieval using Scene Classification in Multimedia Databases

Image and video indexing techniques play a key role in content-based searching in multimedia databases. A novel efficient video similarity search approach for content-based video retrieval in a large storage device is proposed. A compressed domain feature is used to classify the scenes of a video se...

Full description

Saved in:
Bibliographic Details
Published in2021 2nd International Conference for Emerging Technology (INCET) pp. 1 - 7
Main Authors Haloi, Pranabjyoti, Bhuyan, M. K.
Format Conference Proceeding
LanguageEnglish
Published IEEE 21.05.2021
Subjects
and Query by video
Classification algorithms
Entropy
High-dimensional indexing
Image analysis
Key-frame representation
Multimedia databases
Partitioning algorithms
Performance evaluation
Semantics
Video similarity
Online AccessGet full text
DOI10.1109/INCET51464.2021.9456317

Cover

More Information
Summary:Image and video indexing techniques play a key role in content-based searching in multimedia databases. A novel efficient video similarity search approach for content-based video retrieval in a large storage device is proposed. A compressed domain feature is used to classify the scenes of a video sequence. The performance of the said feature in classifying the scenes is evaluated by considering different feature lengths. Subsequently, a novel searching algorithm is proposed based on entropy of the image. This method incorporates temporal information of the video during video retrieval. Additionally, the proposed system has the facility to query both by video and by image. The 92.3% accuracy of the proposed method exceeded that of the VA-file 81.0% and OVA-file 76.4% methods and the computational complexity was lower than the VA-file method. This demonstrates the efficiency and effectiveness of the proposed method.
DOI:10.1109/INCET51464.2021.9456317