We are happy to announce that our paper entitle “Hash-Based Frame Selection for Video Phylogeny” has been accepted at the 8th IEEE International Workshop on Information Forensics and Security (WIFS) 2016, which will take place at the conference Center of NYU Abu Dhabi. The paper is one of the results of a joint work between IC/Unicamp and Deib/Politecnico di Milano.
It deals with the problem of tracing back past history of multimedia documents to discover their ancestral relationships, called Multimedia phylogeny. The authors take into consideration video sequences with different frame rates, which is customary due to video conversion between different standards. Also, they consider that bursts of frames can be deleted anywhere in the stream. The proposed algorithm detects temporal coherent frame pairs among videos exploiting locally sensitive hashing techniques.
The method was validated on a dataset of 2,500 video sequences transformed according to different processing operations. Experiments confirmed that the proposed approach enables frame synchronization in more complex scenarios than those considered up to now in the literature.
A pre-print version of the article is available here.