Analysis on Multimodal Biometric Classification based on Feature Level Fusion and Score Level Fusion Methods

The Multimodal Biometric Classification (MBC) is a process of identifying/verifying a person by combining or fusing a biometric with other biometric traits. In MBC, biometrics like face, iris, fingerprint, vein, gait, voice, etc., are used to recognize the person. However, the limitations in unimodal biometric classification are lower recognition rates, and higher error rates which are solved by the MBC. The MBC based identification and authentication of individuals are performed by Machine Learning (ML) and Deep Learning (DL) approaches. For precise person recognition, the biometric classification methods utilize Feature-Level fusion and Scorelevel fusion techniques for feature fusion of different biometrics. This research represents several classification approaches based on ML and DL methodologies like Support Vector Machine (SVM), K-Nearest Neighbor (KNN), Linear Discriminant Analysis (LDA) and Naive Bayes, Convolutional Neural Network (CNN), and Deep Neural Network (DNN) utilized for MBC of authentication and identification process. This survey examines the strengths, relationships, and challenges of the ML and DL-based classification approaches.