Colour transfer between images

Abstract

Colour transfer is one of the most common tasks in image processing, an efficient method to impose the colour pattern of a video or image on another image. Some of its usage include, but is not limited to, automatic colorization of grayscale images and modifying the impression and atmosphere of a video or image through colour alteration. This report will explore and evaluate two colour transfer methods: the colour transfer method by Reinhard et al [1] which performs colour transference between images using simple statistical analysis of the image colour characteristics; and the N-dimensional probability density function (PDF) transfer [2]. The overall performance of the two algorithms will be assessed using image results and various other factors. Additionally, the report will study the effectiveness of colour transfer when semantic image segmentation using a pre-trained neural network is used.