Color transfer based on color classification

Abstract

In order to solve the problem that the global color transfer algorithm may fail in color-rich images and the resulting images do not have a sense of hierarchy, this paper proposes a color transfer algorithm based on color classification mapping. When faced with a colorful image, the global color transfer algorithm will be weak, and the resulting image will not have a sense of hierarchy. Therefore, it is necessary to use a local algorithm for color transfer. The k-means algorithm based on color classification is used in this paper. Through this algorithm, rich color information can be classified, so that it can be transmitted separately in each color classification. A large number of experimental results show that compared with the global color transfer algorithm, the color transfer algorithm used in this paper is more layered, the result is more accurate, and it has a better performance in terms of visual effects.