A novel method for post-surgery face recognition using sum of facial parts recognition

Abstract

Plastic surgery is becoming more and more commonplace today due to its increasing acceptance in society and its cost-affordability. This in turn has led to the need for developing highly accurate post-surgery face recognition techniques, a problem space which differs significantly from traditional face recognition. In this paper we first conduct a statistical study to show that facial plastic surgery operations correlate with a desire to conform to a golden ratio with respect to the human face. We then apply this knowledge, with the notion of considering a face in terms of the sum of its parts, to propose a novel face recognition technique. The proposed technique is then evaluated against well known datasets, and as per our experiments achieves a recognition rate of 85.35%, which significantly outperforms other state of the art techniques.

Facial Recognition Software Use on Surgically Altered Faces: A Systematic Review

“…While some familiar applications such as facial recognition from photos in social media use these methods, facial tracking using 3-dimensional methods such as infrared scanning for unlocking devices cannot be compared based on these papers. Additionally, some papers mention and even show images of patients having undergone mandible contouring procedures, [16][17][18]22,23,31 but none specifically address the effect of this facial reshaping on the methods employed. Swelling is addressed in two papers reviewing facial changes at different points in time during edema evolution utilizing a specific algorithm that could be employed to identify a face early in the healing period.…”

“…Additionally, some papers mention and even show images of patients having undergone mandible contouring procedures,16–18,22,23,31 but none specifically address the effect of this facial reshaping on the methods employed. Swelling is addressed in two papers reviewing facial changes at different points in time during edema evolution utilizing a specific algorithm that could be employed to identify a face early in the healing period 13,24.…”