Effective and efficient image copy detection based on GPU

Authors:
Hongtao Xie;Ke Gao;Yongdong Zhang;Jintao Li;Yizhi Liu;Huamin Ren
Affiliations:
Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China, Institute of Information Engineering, Chinese Academy of Sciences, Beijing, China;Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China;Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China;Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China;Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China;Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China
Venue:
ECCV'10 Proceedings of the 11th European conference on Trends and Topics in Computer Vision - Volume Part II
Year:
2010

Citing 8
Cited 0

Ordinal Measures for Image Correspondence

IEEE Transactions on Pattern Analysis and Machine Intelligence
The ANN-tree: An Index for Efficient Approximate Nearest Neighbor Search

DASFAA '01 Proceedings of the 7th International Conference on Database Systems for Advanced Applications
Video Google: A Text Retrieval Approach to Object Matching in Videos

ICCV '03 Proceedings of the Ninth IEEE International Conference on Computer Vision - Volume 2
Scale & Affine Invariant Interest Point Detectors

International Journal of Computer Vision
Distinctive Image Features from Scale-Invariant Keypoints

International Journal of Computer Vision
Local invariant feature detectors: a survey

Foundations and Trends® in Computer Graphics and Vision
Hamming Embedding and Weak Geometric Consistency for Large Scale Image Search

ECCV '08 Proceedings of the 10th European Conference on Computer Vision: Part I
SURF: speeded up robust features

ECCV'06 Proceedings of the 9th European conference on Computer Vision - Volume Part I

Quantified Score

Hi-index	0.00

Visualization

Abstract

To improve the accuracy and efficiency of image copy detection, a novel system is proposed based on Graphics Processing Units (GPU). We combine two complementary local features, Harris-Laplace and SURF, to provide a compact representation of an image. By using complementary features, the image is better covered and the detection accuracy becomes less dependent on the actual image content. Moreover, ordinal measure (OM) is applied as semilocal spatial coherent verification. To improve time performance, the process of local features generation and OM calculating are implemented on the GPU through NVIDIA CUDA. Experiments show that our system achieves a 15% precision improvement over the baseline Hamming embedding approach. Compared to the CPU-based method, the GPU realization reaches up to a 30-40x speedup, having real-time performance.