Abstract | ||
---|---|---|
Current studies on salient detection have combined salient cues, such as contrast, prior information, and object edges, which work by segmenting the foreground from the background. However, existing models mostly lose tiny salient regions in scenes and fail to detect all salient regions. In this paper, we propose an optimization algorithm on salient detection based on line sketch. First, we generate the line sketches of images, which include all edges of the salient objects. Second, the detected line sketches are blended into the original images in soft light mode. Then, the sketch works as a mask to highlight the outlines. Finally, the processed images are used in the existing salient detection models. Our contribution is that the salient detection with the sketches extracting image outline can effectively improve the accuracy. Results prove that the proposed algorithm is efficient and improves the performance of models. |
Year | DOI | Venue |
---|---|---|
2020 | 10.1007/s11042-019-08381-8 | Multimedia Tools and Applications |
Keywords | Field | DocType |
Salient detection, Line sketch, Contour, Optimized algorithm | Computer vision,Pattern recognition,Computer science,Computer communication networks,Salient objects,Artificial intelligence,Optimization algorithm,Multimedia information systems,Salient,Sketch | Journal |
Volume | Issue | ISSN |
79 | 9 | 1573-7721 |
Citations | PageRank | References |
0 | 0.34 | 0 |
Authors | ||
4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Ning Li | 1 | 145 | 48.40 |
Hongbo Bi | 2 | 0 | 0.34 |
Hua-Ping Guan | 3 | 3 | 1.74 |
Yulong Li | 4 | 0 | 0.34 |