چكيده به لاتين
Abstract:
Biometric recognition systems especially contacting with twins and their similarities are really important these days. We should think about a biometric system which help us. Although there are so many problems in face with uni-biometric systems, such as noise, not being available, misleading, lack of uniqueness, etc. So usage of multi-biometric systems can be helpful. There are five levels of biometric fusion, one of them which name is score level, it’s so well-known and useful because of its precision and low time complexity, so we preferred it.
Aim of this project is combining two-biometrics of twins at a level of accuracy which is as good as possible .In the proposed method we combine two different discriminant features of biometrics, then combining two biometrics by specific classifiers. At the end we combine three biometrics by this method. Our purpose is increasing accuracy and performance.
We examined proposed method on two different datasets which one of them is our collection. The results are presented as ROC curves and EER and also FRR for constant values of FAR, for all methods being compared. Time complexity is the last part which convince the proposed method are better in comparison with some other methods.
Keywords: biometric, twins, two-biometrics, fusion, authentication, iris, fingerprint, face.
