Fingerprint Identification based on Frequency Domain Analysis

Abstract

This paper concerned with fingerprint authentication based on its class by using effective machine learning such as neural network and Fuzzy C Mean. Gabor filter was used for enhancement fingerprint image. Discrete Wavelet transform was employed as a feature extraction technique in frequency domain. Moreover, Feed Forward Backpropagation Neural network (FFBPNN) and Fuzzy C mean (FCM) were used for classification. To estimate the performance of above two classifiers, a comparison based on recognition efficiency and processing time were obtained. The algorithms were applied on two databases FVC 2004 for 40 persons with eight images per person and another database for 100 persons with ten images per person. The results proved that the FFBPNN is more powerful classification tool in recognition efficiency and execution time. All the programs were executed in MATLAB R 2017b.