CNN Model for Recognition of Human Finger Vein Through Image and Electrocardiogram Signal Analysis

Recently, there's been a buzz about integrating Finger Vein (FV) biometrics with Electrocardiogram (ECG) data, potentially enhancing conventional biometric techniques. Researchers have long explored using ECG signals with FV for human biometric identification despite facing challenges related to generalization and efficiency. To overcome this, we provide a novel approach called Improved Convolutional Neural Networks (I-CNNs) for improving the classification and prediction. Enhancing the efficiency and computation time of the Convolutional Neural Networks (CNNs) was achieved through improvements in convolutional filters, leading to an increased related field of the convolutional layer. We have successfully proved that the improved convolutional model outperforms the classic CNN model by collecting three publicly accessible FV benchmark datasets, namely SDUMLA, FV–USM, and HKPU datasets. These datasets were used for the purpose of testing the suggested technique. Our study confirms that the feasibility of implementing this approach for real-time human authentication using combined FV and ECG signals.