Matching before and after Surgery Faces

Cosmetic surgery is a voluntary procedure usually performed with the motive of improving a person's appearance, removing aging signs and correcting impairments caused due to accidents such as burns or abnormalities. It also serves as a means for frauds and criminals to evade identification. One of the major challenges faced by face recognition algorithms is the effective identification of such altered faces. This paper aims to verify faces before and after facial surgery that will help to seal various security and biometric loopholes also for in general face verification for official records and similar purposes. The paper takes two pictures of a person preferably before and after surgery as input and further on the software removes noise (using interpolation) from the pictures for better quality and then finds fisher vector encodings of the different features (nose, mouth, eye pair) extracted from both the images. The extracted fisher vector encodings are the further passed to a trained classifier which outputs positive or negative indicating matched or unmatched respectively.