Abstract: A new problem in face matching is plastic surgery. Popularity of plastic surgery has increased over the past few years and its keeps growing. This work proposes a facial feature based system to determine the original image from a post-surgery image. The system generates facial components from three different feature extraction techniques. Uniform LBP serves as the classifier Dimensionality reduction techniques such as LDA and PCA are also used. The normalized values are compared and score fusion rule is applied to retrieve the results. On the plastic surgery image portraits, the proposed system yields high recognition accuracy.
Keywords: Face matching, Plastic surgery, ULBP, Forensics.