Name
Abstract | ||
|---|---|---|
We introduce Stable Salient Shapes (SSS), a novel type of affine-covariant regions. The new local features are obtained through a feature-driven detection of Maximally Stable Extremal Regions (MSERs). The feature-driven approach provides alternative domains for MSER detection. Such domains can be viewed as saliency maps in which features related to semantically meaningful structures, e.g., boundaries and symmetry axes, are highlighted and simultaneously delineated under smooth transitions. Compared with MSERs, SSS appear in higher number and are more robust to blur. In addition, SSS are designed to carry most of the image information. Experimental results on a standard benchmark are comparable to the results of state-of-the-art solutions in terms of repeatability score. The computational complexity of the method is also worth of note, as it is lower than those of most of the competing algorithms in the literature. |
Year | DOI | Venue |
|---|---|---|
2012 | 10.1109/DICTA.2012.6411681 | Digital Image Computing Techniques and Applications |
Keywords | Field | DocType |
competitive algorithms,computational complexity,feature extraction,shape recognition,MSER detection,SSS,affine-covariant regions,competing algorithms,computational complexity,feature-driven detection,image information,maximally stable extremal regions,repeatability score,stable salient shapes,symmetry axes | Computer vision,Pattern recognition,Salience (neuroscience),Computer science,Feature extraction,Maximally stable extremal regions,Artificial intelligence,Salient,Computational complexity theory | Conference |
ISBN | Citations | PageRank |
978-1-4673-2179-2 | 2 | 0.38 |
References | Authors | |
15 | 3 | |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Martins, P. | 1 | 2 | 0.38 |
| Paulo Carvalho | 2 | 250 | 47.68 |
| Gatta, C. | 3 | 2 | 0.38 |