The Visual Computer: International Journal of Computer Graphics
| Hi-index | 0.00 |
We present an object-cloning algorithm using constrained mean value interpolation over images. Different from previous methods, we first define the guidance vector field as a weighted gradient of the source and destination images in the gradient domain, which produces a similar Laplace equation to that of the original Poisson method but with different boundary conditions. Then, we use mean value coordinates to solve the new Laplace equation approximately instead of solving a large linear system. The cloned result matches well with the destination image on the luma and preserves the chroma of the source object as much as possible. Our cloned results are visually pleasing without smudging and discoloration artifacts. We also extend our approach to video object cloning. Experimental results demonstrate the effectiveness of our object cloning algorithm.