Abstract:
Fingerprint based authentication systems have developed rapidly in the recent years.
Conventional biometric authentication methods use feature extraction methods and pattern
matching techniques. However, these systems are vulnerable to spoofing attacks. Image
Quality Assessment (IQA) is one of the statistical techniques used in image processing to
determine whether the liveness sample is live or fake. The objective of the proposed system
is to improve the liveness recognition security. In this paper, seven different Full –
Reference (FR) IQA Pixel-wise difference measures are proposed, which are Mean Squared
Error (MSE), Normalized Absolute Error (NAE), Average Difference (AD), Signal to Noise
Ratio (SNR), Peak Signal to Noise Ratio (PSNR), Maximum Difference (MD), and
Structural Content (SC). These seven measures give us seven feature vectors, and then these
feature vectors are classified using SVM classifier. As the image quality measures are used
in the proposed method, the originality of the captured image is preserved as much as
possible. In order to get feature vectors with more features without affecting much the
original image, the unsharp masking (USM) method is used for sharpening the image. The
features extracted using the proposed method are compared with the results from scale invariant feature transform (SIFT), and the comparison reveals that the proposed IQA
method gives much better results than the SIFT.
