Using histogram of oriented gradients for fingerprint classification

This article proposes method for classifying fingerprints. The proposed method can be used for fingerprint identification and biometric analysis. Article describes correlations between fingertip patterns and local area gradient orientation distribution. Method effectiveness is analyzed using two approaches: with simple implementation and with artificial neural network. Theoretical and experimental research proves that the orientation histogram has different characteristics for fingerprints containing different types of patterns. The results of numerical experiments confirm the theoretical conclusions and prove the efficiency of the proposed algorithm.