Optimizing Discriminability of Globally Binarized Face Templates

Biometric systems are being increasingly deployed in various applications such as smartphones, passports and visa control, health cards, and crime investigation applications. Storing individuals biometric templates in the biometric system database makes it susceptible to threats. As a result, many biometric template protection schemes have been proposed in order to provide the protection for the biometric data against unauthorized use. Extracting binary templates from real-valued biometric data is an important stage in biometric template protection systems. Moreover, representing biometric data as binary template can speed up the biometric template processing and reduce the storage capacity needed to store the enrolled templates. Global binarization schemes binarize the original real-valued biometric template using a series of transformation functions applied to the entire template. The main defy of any global binarization scheme is the generation of such transformation functions which optimize the within-class variance and between-class variance for the transformed binary template simultaneously. In this paper, we propose a global biometric data binarization scheme that utilizes genetic algorithm to optimize the within-class variance and between-class variance for the transformed binary template simultaneously, which retains the discriminability of the binary template. Experimental results using face data sets demonstrated promising recognition performance without violating the security needs of the biometric system.