Fake Colorized Image Detection Approaches: A Review

Colorization is a process used in image editing in which grayscale images are colored with realistic colors. Modern techniques of colorization could produce artfully colored images in such a way that it is difficult for human eyes to differentiate between actual and fake colorized images. As a result, identifying fraudulent colored pictures has captured the scientific community’s attention in digital forensics. This paper provides an overview of the strategies used for detecting fake colorized images. Mainly, two approaches were used to design fake colorized image detection systems. The first one uses traditional machine learning (ML) techniques that rely on hand-crafted features derived from images and used to differentiate actual and fake images. The second approach uses deep learning (DL) techniques as “end to end” systems that don’t have to be supplied with such hand-crafted features, as they can learn the features from the image directly. This paper focuses on the various methods and techniques used in fake-colorized image detection. It may aid researchers in better understanding the benefits and drawbacks of existing technologies to develop more efficient systems in this field.