An Overview of Grayscale Image Colorization Methods

摘要

Conversion of grayscale images to color images is a process of adding color to gray, monochrome images in a convincing, visually acceptable way. Nowadays, automated conversion is a challenging area that links machine and deep learning methods with art. Although many experts claim that grayscale images contain a special artistic value, lack of color can be considered as a loss of information. This paper presents an overview of methods and techniques that have been developed for grayscale image colorization. The paper provides a classification of relevant methods, explains the principles on which they are based and emphasizes their advantages and disadvantages. Special focus is put on methods that involve deep learning algorithms. The results show that deep learning colorization methods provide automated conversion and outperform other methods both in terms of quality and speed.