A novel approach to detecting duplicate images using multiple hash tables

This paper presents a new duplicate image detection scheme that adopts multiple hash tables in a novel way for quick image matching and, consequently, fast duplicate detection. The proposed scheme contains two phases: the feature generation phase and the duplication inspection phase. The former phase extracts the features of images that need protection and transforms them into key-value pairs, which are stored in the slots of multiple hash tables. When a possibly duplicated image needs to be examined, the latter phase hashes the features of the suspect image into the corresponding slots of the multiple hash tables and determines if the suspect image is a duplicate one. The execution time of the scheme is relatively short thanks to the unique design of the multiple hash tables. The experimental results show that the proposed scheme obtained satisfactory results both on the recall and precision rates, hence demonstrating it can effectively identify duplicate images including digitally modified copies.