Ordinal Measures for Image Correspondence
IEEE Transactions on Pattern Analysis and Machine Intelligence
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
Distinctive Image Features from Scale-Invariant Keypoints
International Journal of Computer Vision
Scalable Recognition with a Vocabulary Tree
CVPR '06 Proceedings of the 2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Volume 2
Beyond Bags of Features: Spatial Pyramid Matching for Recognizing Natural Scene Categories
CVPR '06 Proceedings of the 2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Volume 2
Evaluation campaigns and TRECVid
MIR '06 Proceedings of the 8th ACM international workshop on Multimedia information retrieval
Hamming Embedding and Weak Geometric Consistency for Large Scale Image Search
ECCV '08 Proceedings of the 10th European Conference on Computer Vision: Part I
Image copy detection using a robust gabor texture descriptor
LS-MMRM '09 Proceedings of the First ACM workshop on Large-scale multimedia retrieval and mining
Video copy detection by fast sequence matching
Proceedings of the ACM International Conference on Image and Video Retrieval
| Hi-index | 0.00 |
Recently, local interest points (also known as key points) are shown to be useful for content based video copy detection. The state-of-art local feature based methods usually build on the bag-of-visual-words model and utilize the inverted index to accelerate search process. In this paper, we offer a detailed description of a novel CBCD system. Compared with the existing local feature based approaches, there are two major differences. First, besides the descriptors, the dominant orientations of local features are also quantized to build the hierarchical inverted index. Second, feature similarity constraints are used to refine the matching of visual words. Experiments performed on a reference video dataset of 50 hours show that our system can deal with 9 types of common video transformations, and due to the hierarchical indexing and feature similarity constraints, the computational costs are reduced as well.