A Deterministic Eviction Model for Removing Redundancies in Video Corpus

Abstrak

The traditional storage approaches are being challenged by huge data volumes. In multimedia content, every file does not necessarily get tagged as an exact duplicate; rather they are prone to editing and resulting in similar copies of the same file. This paper proposes the similarity-based deduplication approach to evict similar duplicates from the archive storage, which compares the samples of binary hashes to identify the duplicates. This eviction is done by initially dividing the query video into dynamic key frames based on the video length. Binary hash codes of these frames are then compared with existing key frames to identify the differences. The similarity score is determined based on these differences, which decides the eradication strategy of duplicate copy. Duplicate elimination goes through two levels, namely removal of exact duplicates and similar duplicates. The proposed approach has shortened the comparison window by comparing only the candidate hash codes based on the dynamic key frames and aims the accurate lossless duplicate removals. The presented work is executed and tested on the produced synthetic video dataset. Results show the reduction in redundant data and increase in the storage space. Binary hashes and similarity scores contributed to achieving good deduplication ratio and overall performance.