Palm vein pattern: Extraction and Authentication

Abstract

The unique approach towards the extraction and authentication of 850 nm near-infrared palm vein pattern is developed in this research. The proposed strategies have amalgamated the predetermined region of interest, preprocessing images, extraction of palm veins pattern, features matching, and user database utilizing the graphical user interface incorporated in Python and OpenCV library. Bank of Gabor filter, which contains 8 Gabor filter in a single bank, is developed along with the Sobel kernel filter to accomplish the desired outcome. Applying the proposed methodology, near-infrared palm vein pattern extraction is efficiently implemented. Radom Forest and Gaussian Naïve Bayes classifiers are used to acquiring the accuracy scores to procure matching. Accuracy scores obtained by the near-infrared palm vein classifier is 97.40% for the Random Forest classifier and 96.30% for the Gaussian Naïve Bayes classifier. The created framework can be joined with different applications, such as record-keeping, interruption identification, qualification confirmation for business exchanges, misrepresentation assessment, and identity verification.