A systematic review of face sketch recognition system / Nur Nabilah Bahrum, Samsul Setumin, Mohd Firdaus Abdullah, Mohd Ikmal Fitri Maruzuki and Adi Izhar Che Ani

Face recognition systems have grown in popularity worldwide due to technological advancements. The use of a sketch to identify a suspect is one of the most common approaches in the field of forensic science when it comes to the application of face recognition algorithms. The difficulties arise when no image capture sources are available. Without an image, it becomes challenging for the police to identify the suspect. In forensic sciences, the identification of face sketch is used when the eyewitness of the crime scene is asked to recall the structures and features of the suspected person, and then the forensic artist will generate a sketch based on the verbal description provided from the eyewitness. Matching these sketches to the corresponding photo is a difficult task. Even though the sketches contain all of the necessary information about the suspect face's spatial topology and geometric details, it is still difficult for manual face recognition to identify and match sketch representations to the corresponding photo. Therefore, an automatic matching method is needed to identify a suspect accurately. This study perform a systematic literature review (SLR) that focuses on discussing of three major aspects of the face sketch recognition system which are the available face sketch database, the recent technique used sketch-photo matching and the contribution of the Generative Adversarial Network (GAN) in face sketch recognition. In this study, 35 different research papers published between 2011 and 2021 are studied and analysed. From this SLR, compared to other methods, GAN had been found to be a great technique for solving a wide variety of general facial recognition problems and also face synthesis.