Human identification using finger vein and ECG signals

Abstract Biometric human identification refers to the automated recognition of individuals based on their biological or behavioral characteristics. Finger vein recognition and electrocardiogram (ECG) recognition have received a lot of attention for over ten years and these two traits are viewed as ones of promising biometric traits because of their unique advantages. However, these two independent unimodal biometrics, which using a single biometric trait for personal recognition, usually cannot satisfy the requirements of real-world applications. Finger vein and ECG signals have their unique advantages and also have their respective shortcomings. It has been well established that multimodal biometrics systems can overcome the respective defects of unimodal approach. Thus, how to jointly exploit these two modalities and develop an effective and efficient multimodal identification system are still open problems. However, to the best of our knowledge, there has been no study on the multimodal biometric systems using finger vein and ECG signals. Therefore, in this paper we make the first attempt to integrate finger vein with ECG signals for personal identification using various fusion strategies. A new multimodal biometric method based on Discriminant correlation analysis (DCA) for the fusion of finger vein and ECG is proposed. Extensive experiments are conducted on a merged bi-modal dataset named VeinECG coming from FVPolyU finger vein dataset and ECG-ID Dataset. Experimental results show that the proposed multimodal system is substantially superior to two individual unimodal systems in terms of both recognition accuracy and security.