REDUCTION OF FALSE ACCEPTANCE RATE IN MULTI-BIOMETRIC SYSTEMS BY WEIGHTED MULTI FEATURE EXTRACTION TECHNIQUE

Abstract

Today, biometrics is most often employed for a variety of mundane activities, such as mobile authentication process and border crossing. Biometrics is subject to strict accuracy and efficiency standards in high-security circumstances. Multi-biometric techniques that combine the data from many biometric elements have shown to reduce erroneous rates and ameliorate the inherent flaws of the separate biometric systems in order to achieve this goal. Thus, to reduce the false acceptance rate in the multi-biometric system, the proposed approach uses the weighted multi-feature extraction technique. In this multi-feature extraction process, the image is initially segmented into multiple parts. Each part is then treated separately for noise reduction and cancellation. Especially the HSI (Hyper-spectral Image) is broken down into multiple Gaussian pyramid for extracting the multiple scaling features and the noise is eliminated by the usage of averaging filter. Further the extracted features are given weights and are featured to form cluster for each and every feature that is being extracted. This method reduces the error rate and provides more efficiency of the system.