Passport Recognition Using Enhanced ART2-based RBF Neural Networks

The judgment of forged passports plays an important role in the immigration control system and requires the automatic recognition of passports as the pre-phase processing. This paper, for the recognition of passports, proposed a novel method using the enhanced RBF network based on ART2. The proposed method extracts code sequence blocks and individual codes by applying the Sobel masking, the smearing and the contour tracking algorithms in turn to passport images. The enhanced RBF network was proposed and used for the recognition of individual codes that applies the ART2 algorithm to the learning structure of the middle layer. The experiment results showed that the proposed method has superior in performance in the recognition of passport.