Safaa Osman, Mehrbakhsh Nilashi, Eko Supriyanto


Face recognition systems and information visualization nowadays are increasingly being used in healthcare for a variety of purposes, including security, patient check-in and check-out, visualization of patient information when they are in a coma or anaesthetic, staff management, and so on. The face pictures obtained for recognition purposes are occasionally subjected to some degradations like low contrast, low resolution, and noise. Over the last few years, Low Resolution Facial Recognition (LRFR) systems have got a lot of attention when high-quality photos are difficult to capture. However, currently, when mandatory and voluntary wearing mask regulations are becoming disperse everywhere during the COVID-19 pandemic in order to prevent the spread of the virus, these circumstances render the traditional facial recognition technologies inadequate because they depend on the features of the whole face. As a result, improving the recognition performance of existing face recognition technology on low-quality masked faces images is critical. To that end, this paper proposes a framework for a system that can recognize masked faces and visualize information and patient or staff identification in healthcare facilities. To achieve this objective, several preprocessing procedures will be performed on the low-resolution masked face photos collection, and the quality will be enhanced before the facial landmarking phase. Thereafter, locating the corners and center of the eyes, geometric shaped of the face, geodetical and Euclidean distances between landmarks, and facial curves will be among the features extracted by this method. Finally, the data will be used to train for the classification objective using Machine Learning techniques. At the end of this research, the system will be able to recognize masked faces and deliver accurate and sensitive identification.

Full Text:



  • There are currently no refbacks.