Multimodal Biometric System Using Hybrid Convolutional Neural Network (HCNN) Based on Face and Finger Vein Fusion

The rising need for information security and security policies worldwide has resulted in the usage of biometric recognition technologies, a common affair in day-to-day life. In lieu of this, multimodal biometrics technology has garnered more focus and popularity owing to its capability of avoiding prominent drawbacks of UBSs (unimodal biometric systems). This work proposes a unique multimodal biometric personal identification system. The suggested system based on DLTs (deep learning techniques) using Hybrid Convolutional Neural Network with Support Vector Machine (HSVM-CNN) can identify people using their facial and finger vein biometrics. In this work, a fully connected layer of CNN will be replaced by a support vector machine. This study's biometric personal identifications fuse facial and finger vein bimodal features. Fusions occur at feature layers of HCNNs (Hybrid Convolution Neural Networks). The user's face and finger biometrics are first collected and analyzed. The fusion processes merge extracted relevant features of the two modalities based on score levels. The experimental results of experiments carried out over available open databases demonstrate enhanced identifications and recognition accuracies with reduced error rates.