Bi-modal palm recognition is an advanced biometric identification system that combines two modalities, visible light palm recognition and palm vein recognition to authenticate and verify individuals. It utilizes the unique characteristics of the palm, including its shape, texture and vein patterns, to provide enhanced accuracy and security.
Palm shape recognition analyzes geometric features such as size, curvature, and finger positions, while palm vein recognition captures the distinct vein patterns beneath the palm’s surface. By integrating these modalities, bi-modal palm recognition systems offer a reliable and contactless solution for applications such as access control and identity verification.
In the training phase, a deep neural network is used to extract features from palm images. These features include shape, texture, reflection, spectrum, and deep semantic features. The network learns to classify live palm images by optimizing the decision boundaries in the feature space. In the testing phase, a stereo camera captures near-infrared palm vein and visible light palm images. Each image is analyzed separately to determine if it is from a live palm. The final liveness probability is obtained by combining the liveness decisions from both modalities using a fusion strategy.
Palm recognition technology is applied in numerous fields, including security, payment, banking, healthcare, and hospitality.From secure access to facilities and devices, to identification for financial transactions, patient verification in healthcare, and personalized guest experiences in hospitality, the applications are broad and burgeoning.
