Issue No. 12 - Dec. (2012 vol. 34)
Mani Malek Esmaeili , Electr. & Comput. Eng. Dept., Univ. of British Columbia, Vancouver, BC, Canada
R. K. Ward , Electr. & Comput. Eng. Dept., Univ. of British Columbia, Vancouver, BC, Canada
M. Fatourechi , Electr. & Comput. Eng. Dept., Univ. of British Columbia, Vancouver, BC, Canada
A fast approximate nearest neighbor search algorithm for the (binary) Hamming space is proposed. The proposed Error Weighted Hashing (EWH) algorithm is up to 20 times faster than the popular locality sensitive hashing (LSH) algorithm and works well even for large nearest neighbor distances where LSH fails. EWH significantly reduces the number of candidate nearest neighbors by weighing them based on the difference between their hash vectors. EWH can be used for multimedia retrieval and copy detection systems that are based on binary fingerprinting. On a fingerprint database with more than 1,000 videos, for a specific detection accuracy, we demonstrate that EWH is more than 10 times faster than LSH. For the same retrieval time, we show that EWH has a significantly better detection accuracy with a 15 times lower error rate.
Nearest neighbor searches, Indexes, Signal processing algorithms, Hamming distance, Algorithm design and analysis, Approximation algorithms, binary embedding, Nearest neighbor search, Hamming space, multimedia fingerprinting, copy retrieval
M. Fatourechi, M. M. Esmaeili and R. K. Ward, "A Fast Approximate Nearest Neighbor Search Algorithm in the Hamming Space," in IEEE Transactions on Pattern Analysis & Machine Intelligence, vol. 34, no. , pp. 2481-2488, 2012.