Robust Fingerprint Minutiae Extraction and Matching using Improved AKAZE

Personal identification is important to protect data from cyber-attacks. Minutiae feature extraction and matching are key tasks for recognizing fingerprints and are essential to Automatic Fingerprint Recognition (AFR) systems. Fingerprint recognition technologies are becoming increasingly significant in preventing identity theft and ensuring accurate identification. However, the major problem is their identification sensors are vulnerable to presentation or spoofing attacks from the fake fingerprint samples. To overcome these problems, an improved AKAZE is introduced for feature extraction and feature matching in the process of fingerprint identification. Thinning and dilation methods are used for preprocessing the image and Improved AKAZE is used for feature extraction of the fingerprint images. After extracting features, the proposed Improved AKAZE is used for identifying and matching those fingerprint images. The proposed method is evaluated on publicly available FVC2004 dataset. The results show that the improved AKAZE achieved 99.95% of accuracy in identification and matching of fingerprint images.