Abstract | ||
---|---|---|
•An effective spectral graph prior is employed to correct the existing priors for obtaining a reasonable high level prior.•A solver of SGLR decomposition is presented for image feature matrix weighted by the high level prior.•Low rank matrix and sparse matrix rather than only sparse matrix are both used in final saliency calculation.•An efficient integration function with an activation function is presented for final saliency generation. |
Year | DOI | Venue |
---|---|---|
2018 | 10.1016/j.jvcir.2017.12.006 | Journal of Visual Communication and Image Representation |
Keywords | Field | DocType |
Saliency detection,Spectral graph,Low rank matrix recovery,Sparse decomposition,Feature matrix | Graph,Computer vision,Salient object detection,Pattern recognition,Salience (neuroscience),Activation function,Salient objects,Low-rank approximation,Artificial intelligence,Prior probability,Mathematics,Sparse matrix | Journal |
Volume | ISSN | Citations |
50 | 1047-3203 | 1 |
PageRank | References | Authors |
0.35 | 41 | 7 |
Name | Order | Citations | PageRank |
---|---|---|---|
Jiazhong Chen | 1 | 33 | 9.85 |
Jie Chen | 2 | 7 | 1.46 |
Hefei Ling | 3 | 241 | 39.63 |
Hua Cao | 4 | 7 | 1.80 |
Weiping Sun | 5 | 14 | 2.66 |
Yebin Fan | 6 | 6 | 1.11 |
Weimin Wu | 7 | 236 | 43.97 |