In recent years, biometric system has used in various applications like surveillance, business areas, security purpose, and etc. so many biometric systemshave been implemented such as vein pattern, iris, retina, Finger Print (FP), and etc. In this work, a fingerprint based biometric system is introduced to improve the system security and the performances. The input image is performed the pre-processing stage such as sharpening, black and white conversion, and resize. Bi-dimensional Empirical Mode Decomposition (BEMD) and SIFT feature extraction are used in this research to create the feature structure image. Convolution Neural Network (CNN) is used for performing the classification which frequently applied to analyze visual imagery. From the classification output, the authentication is given and the performances are evaluated. Finally, the performance parameters are calculated such as Accuracy (A), Sensitivity (SE), Specificity (SP), Precision (P), Recall (R), F measure (FM), G mean (FM), False Rejection Ratio (FRR). The proposed algorithm has achieved 98.74% of accuracy which is more when compared to existing methods.
Volume 11 | 07-Special Issue
Pages: 402-411