This study proposes a novel Multi-Radius Fused Rotational Invariant Uniform Local Binary Pattern variant for finger vein identification, specifically designed for systems with limited computational resources. The proposed operator, \(\textrm{LBP}^{\textrm{riu2}}_{(8,1),(16,1),(8,2)}\) , integrates three distinct radius configurations within a unified sliding window framework to capture multi-scale textural information and enhance discriminative capability. This texture-based feature extraction technique is systematically evaluated in combination with dimensionality reduction methods, including Principal Component Analysis (PCA), Two-Dimensional PCA (2DPCA), and Two-Directional Two-Dimensional PCA ( \(\hbox {2D}^2\) PCA), to assess its robustness across diverse operational scenarios. Comprehensive experiments are conducted on the FV-USM, MMCBNU-6000, and UTFVP datasets. Performance evaluation is carried out using two fusion strategies across three standard experimental protocols. Results demonstrate that the proposed method, along with its dimensionality-reduced variants, achieves competitive performance and outperforms traditional hand-crafted techniques. Furthermore, comparative analyses with the state-of-the-art approaches confirm the effectiveness of the proposed texture-based method. This research establishes the practical performance boundaries of classical feature extraction techniques, with results rigorously validated using Cumulative Match Characteristic (CMC) curves.
