Anti Spoofing Using Convolution Neural Network

EasyChair Preprint no. 7683

Abstract

Biometric technology presents several advantages over classical security methods based on either some information (PIN, Password, etc.) or physical devices (key, card, etc.). However, providing to the sensor a fake physical biometric can be an easy way to overtake the systems security. Fingerprints, in particular, can be easily spoofed from common materials, such as gelatine, silicone, and wood glue. Therefore, a safe fingerprint system must correctly distinguish a spoof from an authentic finger. Different fingerprint liveness detection algorithms have been proposed, and they can be broadly divided into two approaches: hardware and software. The primary purpose of a fingerprint recognition system is to ensure a reliable and accurate user authentication, but the security of the recognition system itself can be jeopardized by spoof attacks. Specifically, we propose a deep convolution neural network-based approach utilizing local patches centered and aligned using fingerprint minutiae. The current experimental results on three public-domains LivDet datasets show that the present approach does not provide the state-of-the-art accuracies in fingerprint spoof detection for intra-sensor, cross- material, cross-sensor, as well as cross-dataset testing scenarios. The earlier fingerprint spoofing system was based on LBP (local binary pattern). However, this does not ensure total security. So, in the new proposed system we are trying to implement the spoof detection using minutiae centered patches. It is supposed that by using this method for fingerprint spoof detection we will observe a significant reduction in the error rates for different spoof attacks.

Keyphrases: ANN Artificial Neural Network, BSIF - Binarized Statistical Image Features, CNN-Convolution Neural Network, LPQ - Local phase Quantization, OCT- Optical Coherence Tomography, RNN - Recurrent Neural Network, WLD - Weber Local Descriptor

@Booklet{EasyChair:7683,
  author = {Avani Khatri and Sakshi Agrawal and Mrs.A.Helen Victoria},
  title = {Anti Spoofing Using Convolution Neural Network},
  howpublished = {EasyChair Preprint no. 7683},

  year = {EasyChair, 2022}}