Fingerprint Binarization using Convex Threshold

A fingerprint identification system is popular and important in biometric identification applications. Its process is composed of two stages: feature extraction and matching. Most fingerprint features include ending points and bifurcation called minutiaes. In this paper, we propose a new fingerprint binarization method based on convex threshold for effective minutiaes extraction. The proposed method is efficient, fast, and simple compared with conventional filtering techniques. Main processes include noise reduction, directional vector estimation, and ridge detection. For noise reduction and directional vector estimation, we use the conventional method. For ridge detection we propose a convex threshold technique using fingerprint structures.