Color transfer between images

Abstract

Image processing has emerged as an important topic in computer vision, enabling a wide range of applications such as image improvement, object detection, and artistic alterations. Color transfer is a fascinating feature of image processing that includes altering the color appearance of an image while keeping its underlying content. Color transfer enhances the creative and aesthetic aspects of image alterations by providing tools for photographers, designers and artists to express their emotions and tell stories through images. The report will be focusing on a comparative analysis of two popular color transfer algorithms: the Reinhard algorithm [1] and the N-Dimensional Probability Color Transfer technique [2]. The usage of image histogram and color clouds provides a quantitative and visual technique to objectively compare the performance of each color transfer algorithm. Furthermore, several other algorithms will be introduced, which includes using Fully Convolutional Networks for Semantic Segmentation [3],to provide further insights on the topic of color transfer.