Abstract | ||
---|---|---|
A novel region-based fusion algorithm to combine infrared (IR) and visible light (ViS) images using a saliency map and interest points is presented. First, the IR image is applied to a saliency detection process to generate a saliency map. Second, the IR image is further analyzed to detect the interest points. Once the freely distributed points of interest have been removed, the salient interest points are extracted. Third, the convex hull of the salient interest points is calculated to obtain the salient region determined by the salient interest points. Then, the initial saliency map is refined by combining the convex hull of the salient interest points. This strategy makes the saliency result more robust and better able to accurately locate the object. Finally, different fusion rules for the object region and the background are employed. Experimental results show that the proposed algorithm has promising performance. |
Year | DOI | Venue |
---|---|---|
2016 | 10.1016/j.neucom.2015.10.080 | Neurocomputing |
Keywords | Field | DocType |
Image fusion,Saliency map,Interest points,Convex hull | Computer vision,Saliency map,Image fusion,Pattern recognition,Salience (neuroscience),Fusion rules,Convex hull,Artificial intelligence,Point of interest,Mathematics,Salient | Journal |
Volume | ISSN | Citations |
177 | 0925-2312 | 6 |
PageRank | References | Authors |
0.43 | 15 | 4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Fanjie Meng | 1 | 23 | 4.93 |
Baolong Guo | 2 | 244 | 32.65 |
Miao Song | 3 | 6 | 0.43 |
Xu Zhang | 4 | 6 | 0.43 |