PZM and DoG based Feature Extraction Technique for Facial Recognition among Monozygotic Twins

Face Recognition of Identical Twin is a challenging task due to the presence of a high degree of correlation in the overall appearance of the face. Few monozygotic twins help with business tricks such as fake insurance compensation. Most importantly, if one of the indistinguishable twins commits a serious crime, their unclear personalities cause confusion and uncertainty in court trials. The proposed method can be employed for these applications to overcome such harms. In this paper, The AdaBoost Technique is employed for the face detection using Haar features. This algorithm identifies the face region of the input image. The Pseudo Zernike Moment (PZM) and Difference of Gaussian (DoG) methods are utilized to extract the features from the face region detected by AdaBoost algorithm and stored in the databases in both training and testing phase. The Support Vector Machine (SVM) classifier distinguishes the twin's features by comparing both trained and tested features and identifies the culprit who is required as a result. The experimental results demonstrated the ability of the proposed method to recognize a pair of Identical twins.