Automatic color transfer based on clustering
Changliang Wang · Computer Engineering and Applications Journal · 2007
A new clustering-based algorithm for color transfer is proposed in this paper.Global or local color transfer between color images or between color image and gray-scale image can be performed by this algorithm.Firstly,GMM-EM algorithm is jointly applied to automatically group color source image into regions.The same GMM-EM algorithm is used if target image is color image and k-means algorithm is used in case of gray-scale image.Then,corresponding map will be built between source image regions and target image regions after clustering.Finally,Lαβ values of pixel in the target image are changed to impose domain color of source image on target image.The experiments show the presented algorithm is highly automatically and effective and,satisfying results can be obtained.