An Analysis of Fingerprint Identification using VIT-TEXPAD Techniques

An unified fingerprint identification framework that combines a Vision Transformer (ViT) backbone with texture‑patch–based descriptors (TexPAD) for joint matching and presentation attack detection (PAD). The proposed ViT‑TexPAD model uses local patch embeddings from ViT to encode fine‑scale ridge and pore texture, improving both genuine user recognition accuracy and spoof detection robustness. On benchmark datasets similar to LivDet 2013/2015, ViT‑TexPAD achieves higher integrated matching accuracy and lower equal error rate (EER) than CNN baselines and vanilla ViT recognition, while using a comparable or reduced number of parameters. These findings demonstrate that transformer‑based local texture patches constitute a powerful representation for modern fingerprint identification systems