Color transfer based on multiscale gradient-aware decomposition and color distribution mapping
Zhuo Su, Daiguo Deng, Xue Yang, Xiaonan Luo · 2012
Automatic global color transfer is a challenging problem in image editing. In this paper, we propose a novel color transfer method, which is based on the gradient-aware decomposition and the color distribution mapping. Firstly, a gradient-aware decomposition model is established to separate the target image into the base and detail layers. Then, the colors of each separated base layer are enforced to match those of a given reference image by Pitie's multi-dimensional probability density function transfer method. After that, all mapped base layers are combined with corresponding boosted detail layers to produce the final output. The experiments demonstrate that our method can achieve a visual satisfied result without post-processing gradient correction.