Name
Abstract | ||
|---|---|---|
Local features are widely used for content-based image retrieval and object recognition. Most feature descriptors are calculated from the gradients of a canonical patch around repeatable keypoints in the image. In this paper, we propose a technique for designing quantization matrices that reduce the mean squared error distortion of the gradient derived from DCT-encoded canonical patches. Experimental results demonstrate that our proposed patch encoder greatly outperforms a JPEG encoder at the same encoding complexity. Moreover, our quantization matrices achieve lower gradient distortion and larger number of feature matches at the same bit-rate. |
Year | DOI | Venue |
|---|---|---|
2012 | 10.1109/ICIP.2012.6467407 | ICIP |
Keywords | Field | DocType |
gradient,image coding,image matching,gradient distortion,feature descriptors,mean squared error distortion,image compression,matrix algebra,quantization,discrete cosine transforms,gradient preserving quantization,content-based image retrieval,encoding complexity,object recognition,local feature,dct-encoded canonical patch,quantization matrix,mean square error methods | Computer vision,Feature detection (computer vision),Pattern recognition,Computer science,Matrix (mathematics),Image retrieval,Artificial intelligence,Encoder,Quantization (image processing),Quantization (signal processing),Distortion,Cognitive neuroscience of visual object recognition | Conference |
ISSN | ISBN | Citations |
1522-4880 E-ISBN : 978-1-4673-2532-5 | 978-1-4673-2532-5 | 2 |
PageRank | References | Authors |
0.38 | 6 | 4 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Mina Makar | 1 | 93 | 7.27 |
| Haricharan Lakshman | 2 | 328 | 30.58 |
| Vijay Chandrasekhar | 3 | 949 | 45.35 |
| Bernd Girod | 4 | 8988 | 1062.96 |