A Proposed Grayscale Face Image Colorization System using Particle Swarm Optimization

Introduction

Color information is a key attribute of an image which finds extensive uses in various important fields like medical image processing, facial image processing, satellite image processing, video processing etc. (Umbaugh,1998, Gonzalez, 2001, Gonzalez, 2010). Colorization is process to convert a grayscale image into a color one and it is a long sought after goal in field of image processing (Zhang, 2016). Colorization has a wide application in the area of archaeology, medical application, entertainment, law enforcement etc. In archaeology, the gray scale archive documents can be preserved by enhancing it into a colorized one. In entertainment, manual colorization of old movies and video clips (grayscale model) is normally a laborious process (Zhang, 2016). Moreover, a color face contains information like complexion, hair color, eye ball color etc. which are required for searching a face image from color face database and hence colorization of grayscale face image is important. As the gray scale face images do not contain any color information, no color based face image processing algorithms (Hadid, 2002, Bhattacharjee, 2009, Chai, 1999) can be applied on them. Our goal is not to recover the actual ground truth color but to convert the grayscale image into a visually compelling color image so that it looks natural.