Chapter 10 - Local plastic surgery-based face recognition using convolutional neural networks

Abstract

Face recognition has established itself as a widely used application in various areas of social networks such as Facebook, Instagram, Twitter, etc. However, an emerging issue that affects the performance of face recognition is facial plastic surgery, which occurs as a result of medical surgical procedure. We have addressed local plastic surgery face recognition with three different types of local facial plastic surgeries that change certain portions of the face permanently: rhinoplasty, blepharoplasty and lip augmentation. Furthermore, we have proposed local plastic surgery -based face recognition using a convolutional neural network (CNN). It automatically extracts high-level features, unlike the hand-crafted features as in state-of-the-art face recognition algorithms. The proposed nine-layer CNN model local plastic surgery -based face recognition has been evaluated on the Plastic Surgery Face Database (PSD) and the American Society of Plastic Surgeons Face Database (ASPS), which consist of presurgical face images and postsurgical face images of variable sizes. We have performed experimentation with these databases and achieved accuracy ranges from 90%–100%. From the result, CNN outperformed in terms of accuracy as compared to state-of-the-art face recognition algorithms, irrespective of image size and with increasing sizes of databases.