| Mass Transport Based Multi-Model Color Transfer |
| Received:October 28, 2016 Revised:December 19, 2016 |
| Key Words:
Mass transport color transfer histogram CIELab color space
|
| Fund Project: |
| Author Name | Affiliation | | Yanto LIU | School of Mathematical Sciences, Dalian University of Technology, Liaoning 116024, P. R. China
School of Science, Dalian Minzu University, Liaoning 116600, P. R. China | | Shengjing TIAN | School of Mathematical Sciences, Dalian University of Technology, Liaoning 116024, P. R. China | | Meng LIU | School of Mathematical Sciences, Dalian University of Technology, Liaoning 116024, P. R. China | | Xiuping LIU | School of Mathematical Sciences, Dalian University of Technology, Liaoning 116024, P. R. China |
|
| Hits: 3168 |
| Download times: 1518 |
| Abstract: |
| Image editing is a hot research topic in the image processing region, and one of the most basic problems in this field is the color transfer. It modifies the color style of the target image to match the color palette of the source while preserving the content of the target. Although there exist many algorithms to deal with this problem, none of existing methods can handle all kinds of images, especially complex landscape images. In this paper, we propose a Mass Transport based color transfer method in the CIELab color space, which can generate natural transfer results for different kinds of images including landscapes. Different from previous methods using the same transfer model for all the channels, we adopt different models for the luminance channel and color channels. For the luminance channel, a global transfer method is adopted for keeping the consistence in the visual perception. For the color channels, we construct a model based on histogram and Mass Transport, which perfectly transfers the colors of all the pixels from the source to the target and simultaneously preserves the content. What is more, experiments on real images demonstrate that our approach performs favorably against most of the existing methods. |
| Citation: |
| DOI:10.3770/j.issn:2095-2651.2017.01.011 |
| View Full Text View/Add Comment |
|
|
|
