Graph Based Video Sequence Matching & BoF Method for Video Copy detection

Abstract

In this paper we propose video copy detection method using Bag-of-Features and showing acyclic graph of matching frames of videos. This include use of both local (line, texture, color) and global (Scale Invariant Feature Transform i.e. SIFT) features. This process includes dividing video into small frames using dual threshold method which eliminates the redundant frames and select unique key frames. After that from each key frame binary features are extracted which known as Bag of Features (BoF) which are get stored into the database in format of matrix. When any query video is being uploading, same features are extracted and compared with stored database to detect copied video. If video detected as copied then using Graph Based Sequence Matching Method, actual matched sequence between key frames is displayed in acyclic graph.
DOI: 10.17762/ijritcc2321-8169.150673