The Research On Color Transfer For Clothing-color Changing

Color transfer is a subject, which has a great significance to theoretical and practical value of computer vision. So it attracts many researchers who has done many meaningful work. But the existing color transfer algorithm is to transfer the color by the whole proportion, while it can not adjust the source image and the target image on the final composite image proportion of dedication. Under the circumstance of the difference between the two images is too large, it will cause the excessive color transfer and the phenomenon of the discordant image. In the view of this problem, we researched the color transfer algorithms in this article. And we applied this improved color transfer algorithm to image processing of clothing color-changing. It has broaden the color selection of color-changing and has given a transition to the edge, so it has given people a better visual sense effect.The paper mainly includes the following aspects:Firstly, we introduced the significance and the status of this topic. And then we expounded methods of the color transfer and had a brief introduction of the main work in this paper.Secondly, after the in-depth analysis of the Appropriate selection of color space and the color transfer algorithm based on the features of lower order moments and higher order value moment, we put forward an improved color transfer algorithm based on statistical information. Taken into account almost all the existing color transfer algorithms focus on the full-scale image transmission which will cause the imbalance phenomenon, the algorithm in this paper selected the optimum transfer ratio automatically.And it is effective to solve this problem.Finally, for the limitations of the color selection which in traditional clothing-color changing, we applied the color transfer algorithm to the clothing-color change. In this way, the color selection is not limited to the Pantone color choices, it also be selected based on the user’s favorite color image. And we dealt with the edge which is not smooth after the color transfer. After the processing, the edge is smoother and looks more natural.