Speed up duplicate/near-duplicate image detection

Authors:
Chunlei Yang;Jinye Peng;Xiaoyi Feng;Jianping Fan
Affiliations:
Northwestern Polytechnical University, Xian, P.R. China and Charlotte, NC;Northwestern Polytechnical University, Xian, P.R. China;Northwestern Polytechnical University, Xian, P.R. China;Northwestern Polytechnical University, Xian, P.R. China and Charlotte, NC
Venue:
ICIMCS '10 Proceedings of the Second International Conference on Internet Multimedia Computing and Service
Year:
2010

Citing 7
Cited 0

Approximate nearest neighbors: towards removing the curse of dimensionality

STOC '98 Proceedings of the thirtieth annual ACM symposium on Theory of computing
Multi-scale sub-image search

MULTIMEDIA '99 Proceedings of the seventh ACM international conference on Multimedia (Part 2)
Detecting image near-duplicate by stochastic attributed relational graph matching with learning

Proceedings of the 12th annual ACM international conference on Multimedia
Speeded-Up Robust Features (SURF)

Computer Vision and Image Understanding
Large scale image copy detection evaluation

MIR '08 Proceedings of the 1st ACM international conference on Multimedia information retrieval
Near-duplicate detection for images and videos

LS-MMRM '09 Proceedings of the First ACM workshop on Large-scale multimedia retrieval and mining
Real-time near-duplicate elimination for web video search with content and context

IEEE Transactions on Multimedia - Special issue on integration of context and content

Quantified Score

Hi-index	0.00

Visualization

Abstract

Finding duplicate and near-duplicate images plays an important role on redundancy reduction for image storage, summarization and recommendation. This paper introduces how to speed up Duplicate/Near-Duplicate(D/ND) image detection. Image clustering was first applied to partition the images into multiple groups by using coarse visual features; pair-wise image matching was further applied on the images within the same cluster by using fine visual features such as interesting point descriptors. Our coarse-to-fine method can dramatically reduce the computation cost while achieving comparable detection accuracy rate.