A New Cancelable Deep Biometric Feature Using Chaotic Maps

In addition to the high security required for sensitive data in recent years, the availability of low-cost data acquisition devices for biometrics and impressive advances in digital technology have significantly increased the use of biometric technologies for automatic human identity recognition. Unfortunately, human biometrics are very sensitive due to constant communication with consumers. This justifies growing concerns about human integrity and anonymity before any hacking attempt. Therefore, much research has been focused on extracting reversible biometric functions and finding a way to replace them whenever they are compromised. In this paper, a new cancelable deep feature extraction method (C-PCANet) using chaotic maps is proposed. Our scheme can effectively provide lightweight and cancelable deep biometric features that can employed in a variety of high-security applications.