It is obvious that tampering of raw biometric samples is becoming an important security and privacy concern. The Benford's law, which is also called the first digit law, has been reported in the forensic literature to be very effective in detecting forged or tampered data. In this chapter, besides an introduction to the concept and state-ofthe-art reviews, the divergence values of Benford's law are used as input features for a neural network for the classification of biometric images. Experimental analysis shows that the classification of the biometric images can achieve good accuracies between the range of 90.02% and 100%.

Chapter Contents:

• 11.1 Introduction
• 11.2 Related work
• 11.2.1 Benford's law
• 11.2.2 Neural networks
• 11.2.3 Mixed biometric data classification to preserve privacy
• 11.3 Experimental setup
• 11.3.1 Separation of different types of biometric images
• 11.3.2 Data sets
• 11.3.3 Divergence metric and separability of biometric databases
• 11.4 Proposed method
• 11.4.1 Method description
• 11.5 Results and discussion
• 11.5.1 Inter-class separability of biometric images
• 11.5.2 Intra-class separability of biometric images
• 11.5.3 Mixed inter-class and intra-class separability of biometric images
• 11.5.4 Comparative analysis between inter-class, intra-class and mixture of inter-class and intra-class classification of biometric images
• 11.6 Conclusion and future work
• References

Inspec keywords: image classification; digital forensics; neural nets; image coding; statistical distributions; biometrics (access control)

Other keywords: biometric samples tampering; Benford law; forensic literature; biometric images classification; neural network

Subjects: Data security; Computer vision and image processing techniques; Other topics in statistics; Other topics in statistics; Neural computing techniques; Image recognition; Image and video coding