Abstract: For color transfer between colorful images, we used K-means clustering to classify image pixels and we also proposed nearest region matching algorithm. This algorithm can avoid the problem that multiple regions matched to the same region, and it can get the best match results between the two images. Translate the color space from RGB to for the two images. Divide color image and luminance mapped shape image using K-means models into the same number of classes. Determine the relationship between regions of the two images using Euclidean distance and the nearest region matching algorithm. Complete color transfer from color image to the shape image. Experimental results show that the algorithm can achieve the color transfer better for colorful images.

Key words: K-means clustering, region matching, color transfer, colorful image