.png){kind=icon}
.png){kind=icon}
Return
Image copy-and-paste with optimized gradient
.png){kind=icon}
PRE
AI
DOI：10.1007/s00371-013-0897-y.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
This paper presents a novel approach to image copy-and-paste with optimized gradient. We improve the traditional gradient domain cloning by creating smooth transition from source patches to target images. We first specify a source patch and select its foreground region by strokes. Next, we create a gradient transition map in the cloning area. Finally, we reconstruct the gradient of the source patch according to the gradient transition map, and propose an interpolation based method to efficiently calculate the composition results, which avoids solving a large linear system. Experimental results demonstrate the effectiveness of our method, which can produce more natural and satisfying results.
Keywords:
Copy-and-paste
Gradient transition map
Smooth transition
Image cloning
Journal
IF:
2.9
Papers:
4.5K
Citations:
6.5K
