Multimodal Biometric Human Recognition System—A Convolution Neural Network based Approach
Dharmendra Kumar, Sudhansh Sharma, Mangala Prasad Mishra · 2022
In this advance era of digitalization, the trend towards the automation of recognition of individual's identity based upon physiological or behavioural characteristics has become the necessity & need of the time. Various, research have been conducted on multimodal identification biometric system with three traits like face, IRIS & finger vein traits and the model performance was evaluated on score level fusion, feature level fusion and at the accuracy level. However, for better results on identity recognition, it is required to fuse more traits and hence to study the improvements achieved, while recognizing any identity in person. The multimodal biometric fusion of more features supports to create robust system for identity identification. The performed work intends to explore the effect of using deep learning algorithms on identity identification by using a wider range of recognition traits like iris, face, and fingerprint & handwritten signature. The results of the performed work shows that fusion of handwritten signature along with the three traits like face, IRIS & finger vein, leads to a much robust multimodal biometric feature fusion which leads to considerable improvement in the physical identification of any person.