Fingerprint Liveliness Detection using Stacked Ensemble and Transfer Learning Technique

The effectiveness and dependability of fingerprint recognition systems make them popular for biometric authentication. This research addresses the challenge of distinguishing between real and spoof fingerprint images using a comprehensive approach that integrates data preprocessing, data augmentation, feature extraction with pretrained models and stacked classifiers. The study begins with rigorous data preprocessing techniques to enhance the quality and consistency of the fingerprint images. Subsequently, data augmentation is employed to increase the diversity of the training dataset, enabling better generalization and robustness of the classification model. Feature extraction is then performed using pretrained convolutional neural network models, namely ResNet50, VGG16, and DenseNet201, to capture discriminative features from the fingerprint images.Stacked classifiers leverage the complementary strengths of multiple base classifiers, such as Random Forest, Extra Trees, and Support Vector Classifier, to improve classification performance. The proposed ensemble learning approach achieves an impressive accuracy of $\mathbf{9 9 . 3 0 \%}$ in distinguishing between real and spoof fingerprint images, demonstrating its effectiveness in enhancing biometric security systems.