Pattern Recognition of Surgically Altered Face Images Using Multi-Objective Evolutionary Algorithm

Abstract

Plastic surgery has been recently coming up with a new and important aspect of face recognition alongside pose, expression, illumination, aging and disguise. Plastic surgery procedures changes the texture, appearance and the shape of different facial regions. Therefore, it is difficult for conventional face recognition algorithms to match a post-surgery face image with a pre-surgery face image. The non-linear variations produced by plastic surgery procedures are hard to be addressed using current face recognition algorithms. The multi-objective evolutionary algorithm is a novel approach for pattern recognition of surgically altered face images. The algorithms starts with generating non-disjoint face granules and two feature extractors EUCLBP (Extended Uniform Circular Local Binary Pattern) and SIFT (Scale Invariant Feature Transform), are used to extract discriminating facial information from face granules.
DOI: 10.17762/ijritcc2321-8169.1503162