Palm-Vein Recognition Using RGB Images
Sungchul Cho, Kar‐Ann Toh · 2018
This paper addresses the problem of palm-vein recognition based on the RGB color images. Comparing with existing palm-vein methods based on the infrared (IR) or near-infrared (NIR) images, this is the first study of palm-vein recognition utilizing the RGB images. The proposed system consists of three processing steps namely, image enhancement, vein region detection, and feature matching. Essentially, the image enhancement step emphasizes the contrast of the palm-veins for visibility. The lines of the palm-veins are subsequently detected using a simplified Gabor filter which is followed by a binarization for noise removal. Finally, the extracted binary templates are matched using the hamming distance. The system shows an equal error rate of 0.87% on the publicly available PolyU database which consists of four channels: Red, Green, Blue and NIR. This result establishes the feasibility of using the RGB images for palm-vein recognition.