Design of fake fingerprint detection system using enhanced machine learning-based classifier

Fingerprint authentication is commonly utilized in the biometrics sector because of the convenience with which it can be performed; yet, it is susceptible to being misled by fake fingerprints. The fake occurrence can potentially estimate the degree of life remaining on a finger. It may not be appropriate for real-time authentications, however, since it takes two consecutive fingerprints to detect a fake. To solve these issues, we suggest an enhanced Naive Bayes-based Support vector machine (ENB-SVM) that only requires one fingerprint to determine liveness. Principal component analysis (PCA) uses the dimensionalities of feature extraction. A novel liveness detection method functions extremely well in experiments, using just one fingerprint and no additional hardware to determine vitality. The suggested approach demonstrated a fake fingerprint detection method with an average classification error rate of 1.20%.