Performance analysis of ensembled classifier methods for human multimodal biometric recognition

Abstract

Biometrics recognition is the progression of determining an individual s activities and biological characteristics Biometric systems are used increasingly to distinguish human individuals and to access the information about the biometric behaviour Biometric recognition comprises of various stages such as feature extraction similar feature matching relevant feature estimation and collection of extracted features Here feature matching process is developed when there are variants in biological attributes and behaviors surrounded among various persons The biometric systems are mostly employed to overcome the challenging issues on security Thus it protects the individual human information with efficient recognition In recent times many investigations have been considered for attaining enhanced biometric recognition In existing multimodal biometric algorithm relevant features were not extracted due to the presence of high dimensional features and hence the recognition rate needs to be compromised Similarly various conventional techniques were considered for better feature extraction by removing unwanted noisy images In addition due to performance degradation recognition rate was said to be compromised for multimodal processes and hence made the biometric recognition system extremely complex However multimodal biometric system integrates several sources of biometrics information to form more genuine recognition In order to overcome above such issues like performance degradation presence of high dimensional features removing unwanted noisy images three different techniques such as Ensembled Support Vector Machine based Kernel Mapping ESVM KM technique Deep Contourlet Derivative Weighted Rank DCD WR framework and Geometric Curvelet and Minkowski Multimodal Biometric Recognition GC MMBR method are developed with various optimal feature selection process and SVM classifier in multimodal biometric recognition newline