Name
Abstract | ||
|---|---|---|
We present a method to extract salient regions in natural images based on local extrema. Behavioral and physiological studies on human vision indicate that the peripheral vision has an important role to get salient regions into the fovea to obtain detailed information from the environment. Modeling of the peripheral is the key to create a salient region extraction method with the behavioral plausibility. To extract salient regions in a natural image, we focus on the multiresolutional 2-D distribution of local extrema. Local extrema detect regions in the image that are either brighter or darker than the surroundings. They will provide useful information to extract salient regions regardless of the image resolution. As for the function of human vision, an image is more blurred in more peripheral. So we model the peripheral using scale-space image representation. In this study, we define the saliency based on the stability of local extrema on scale-space and create a method to extract salient regions. Comparing the human map of fixations, we ensure that our method successfully extracts salient regions. |
Year | DOI | Venue |
|---|---|---|
2010 | 10.1109/ICIP.2010.5649294 | Image Processing |
Keywords | Field | DocType |
computer vision,feature extraction,image representation,image resolution,natural scenes,object detection,behavioral plausibility,human vision,image resolution,local extrema,multiresolutional 2D distribution,natural image,peripheral vision,region detection,salient region extraction,scale-space image representation,local extrema,peripheral vision model,salient region | Computer vision,Object detection,Pattern recognition,Salience (neuroscience),Visualization,Computer science,Feature extraction,Maxima and minima,Peripheral vision,Artificial intelligence,Image resolution,Salient | Conference |
ISSN | ISBN | Citations |
1522-4880 E-ISBN : 978-1-4244-7993-1 | 978-1-4244-7993-1 | 1 |
PageRank | References | Authors |
0.36 | 5 | 3 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Hidenori Maruta | 1 | 16 | 7.11 |
| Masahiro Ishii | 2 | 1 | 0.36 |
| Makoto Sato | 3 | 60 | 8.47 |