Deep transformer-enhanced dynamic structured light for three-dimensional moving object imaging
This work combines transformer-based fringe-image correspondence tracking with dynamic structured light, estimating pixel-wise motion-induced phase errors and compensating PSP reconstruction for accurate 3D imaging of moving objects.
Dec 15, 2025
VeinTr: Robust end-to-end full-hand vein identification with transformer
VeinTr is an end-to-end transformer model for full-hand vein biometric identification that captures global vein topology through self-attention, achieving robust recognition across imaging conditions and hand poses without manual feature engineering.
Oct 1, 2024
TopFormer: Topology-Aware Transformer for Point Cloud Registration
TopFormer augments transformer-based point cloud registration with topological structure descriptors, improving correspondence quality in geometrically repetitive and low-texture regions by encoding how local surface neighbourhoods connect globally.
Apr 1, 2024