Gray Scale Image Colorization for Human Faces

Image colorization is an eclectic topic of research. Former approaches to the gray scale image colorization problem rely on manual methods with human intervention that produced some de-saturated results that are not likely to be true colorizations. To fully automate the process of colorization, enormous amount of computation power and datasets are required which is daunting. This paper proposes an approach to fully automate the task using an autoencoder model. This can be useful in colorizing old black and white pictures and movie snapshots. We used Convolution Neural Network (CNN) for colorization because of its ability to deal with image datasets. The experimental results demonstrated that this method outperforms the state of art technique.