Edge-based Representation and Recognition for Surgically Altered Face Images

Basically, plastic surgery procedure introduces skin texture variations between images of the same person (intra-face) thereby making recognition more difficult than in normal scenario. Since the shape of significant face features such as eyes, nose, eyebrow and mouth remains unchanged even after plastic surgery, edge-based recognition methods can be employed. This paper employs an edge-based Gabor feature representation approach for the recognition of surgically altered faces. We use the edge information, which is dependent on the shapes of the significant face features, to address the problems of texture variations due to plastic surgery procedures. To ensure that the edge information richly captures significant features of the face and with little or no false edges, a simple illumination normalization step(s) is proposed prior to edge information extraction. Then, the Gabor wavelet is applied on the edge image, which accentuates on the uniqueness of the significant features for discriminating amongst different persons. Experimental results on plastic surgery database (Rhytidectomy, Rhinoplasty and Blepharoplasty) shows that the proposed method performs significantly well in comparison to existing plastic surgery, face recognition methods reported in the literature.