New Hierarchical Finger-Vein Feature Extraction Method for iVehicles

With the advancement of multimedia and digital technology, traditional vehicles are gradually replaced by intelligent ones. As people attach increasing importance to convenience and security, traditional keys and password locks are also being replaced. Although radio frequency identification (RFID) is convenient, some researches have pointed out security concerns on its unlocking technology. In view of this, the finger-vein patterns to be used as a keyless vehicle access control system for intelligent vehicles (iVehicles) is presented. Semantic segmentation DeepLabv
$3^{+}$
based on deep learning (DL) was integrated to filter out the background noise and enhance processing stability. Also, the enhanced maximum curvature (EMC) method to extract vein features was adopted, and best matching regional scores (SMRS) and support vector machines (SVMs) were utilized for hierarchical feature extraction. Lastly, these methods were actualized on a low-level embedded platform Raspberry Pi, with which cloud computing was used to realize real-time identification. When three images were used for training and three for testing, the results showed that the proposed hierarchical vein verification technique had an equal error rate (EER) of 0.84% and 0.47% in the NIU-MIT and FV-USM datasets, respectively.