🤖 AI Summary
To address presentation attack threats—both known and unknown materials—in fingerprint recognition systems, this paper proposes a liveness detection method that dynamically fuses deep convolutional neural network (CNN) features with handcrafted features (LBP and Gabor). The method introduces an adaptive dynamic weighting fusion mechanism that synergistically leverages the high discriminability of CNNs and the interpretability and robustness of handcrafted features, thereby overcoming the limitations of single-feature paradigms. Evaluated on the multi-protocol LivDet 2015, 2017, and 2019 benchmarks, the approach achieves overall accuracies of 96.10%, 96.49%, and 94.99%, respectively—significantly outperforming state-of-the-art methods at the time. The results demonstrate strong cross-material and cross-sensor generalization capability, enabling robust liveness detection in real-world deployment scenarios.
📝 Abstract
Automatic fingerprint recognition systems suffer from the threat of presentation attacks due to their wide range of deployment in areas including national borders and commercial applications. A presentation attack can be performed by creating a spoof of a user's fingerprint with or without their consent. This paper presents a dynamic ensemble of deep CNN and handcrafted features to detect presentation attacks in known-material and unknown-material protocols of the liveness detection competition. The proposed presentation attack detection model, in this way, utilizes the capabilities of both deep CNN and handcrafted features techniques and exhibits better performance than their individual performances. We have validated our proposed method on benchmark databases from the Liveness Detection Competition in 2015, 2017, and 2019, yielding overall accuracy of 96.10%, 96.49%, and 94.99% on them, respectively. The proposed method outperforms state-of-the-art methods in terms of classification accuracy.