Abstract | ||
---|---|---|
In this letter, we introduce a new image magnification method based on subpixel estimation. Different from general methods, we consider each pixel in an RGB color image as a linear combination of distinct object classes, and decompose it to an appropriate set of subpixels. This is done through a local search in a lookup table to get best matches, considering the pixel intensity level and its neighboring structure. The lookup table is obtained from input image which contains same target classes. This eliminates the need of any other training sets. The presented method is simple, capable of online implementation and preserves edges better. The results also show better visual effects and objective measures of quality and sharpness compared to interpolation methods. |
Year | DOI | Venue |
---|---|---|
2014 | 10.1109/LSP.2014.2311957 | Signal Processing Letters, IEEE |
Keywords | Field | DocType |
image classification,image colour analysis,interpolation,table lookup,RGB color image,distinct object classes,image magnification method,interpolation method,lookup table,pixel intensity level,subpixel decomposition,subpixel estimation,target classes,Image magnification,lookup table,subpixel decomposition | Computer vision,Lookup table,Pattern recognition,Demosaicing,RGB color model,Pixel,Artificial intelligence,Subpixel rendering,Contextual image classification,3D lookup table,Mathematics,Color image | Journal |
Volume | Issue | ISSN |
21 | 5 | 1070-9908 |
Citations | PageRank | References |
0 | 0.34 | 13 |
Authors | ||
3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Mansoor Zeinali | 1 | 0 | 0.34 |
Hassan Ghassemian | 2 | 396 | 34.04 |
Mohammad Naser-Moghaddasi | 3 | 0 | 0.34 |