This dissertation by Abhishek Nagar, submitted to Michigan State University for the degree of Doctor of Philosophy in Computer Science, focuses on the security of biometric templates. It highlights the increasing vulnerability of biometric recognition systems due to the proliferation of such systems and the potential for system compromise. The dissertation provides a comprehensive analysis of the vulnerabilities in biometric recognition systems, particularly those related to the information stored in biometric templates. It demonstrates that fingerprint images can be recovered from Minutiae Cylinder Codes (MCC), a well-known fingerprint representation, with high accuracy, which can then be used to create spoof fingers and compromise enrolled systems. The dissertation categorizes techniques to protect biometric templates into two main groups: biometric cryptosystems and template transformation techniques. Biometric cryptosystems involve binding a secure key to biometric data to create a secure sketch from which no information about the biometric data or the key can be recovered. Template transformation techniques, on the other hand, non-invertibly transform the biometric template with the user's password. The dissertation studies two main examples of biometric cryptosystems: fuzzy vault and fuzzy commitment, and provides an improved security analysis that accounts for the non-uniform distribution of biometric features. It also proposes a framework to effectively combine multiple biometric representations. The dissertation identifies two limitations of typical biometric cryptosystems: linkability, which allows identifying two secure biometric templates generated from the same biometric, and the use of only simple biometric representations. It develops techniques to overcome these limitations in the context of fuzzy vault. Additionally, it evaluates various template transformation techniques proposed in the literature, assessing their security using comprehensive metrics. The analysis of template inversion, or the recovery of the original template from a transformed template, is a key element of the security analysis. The dissertation emphasizes the critical importance of protecting biometric templates for public acceptability and user privacy. It believes that the security analysis presented will help streamline the development of new techniques and find robust solutions for protecting biometric data.This dissertation by Abhishek Nagar, submitted to Michigan State University for the degree of Doctor of Philosophy in Computer Science, focuses on the security of biometric templates. It highlights the increasing vulnerability of biometric recognition systems due to the proliferation of such systems and the potential for system compromise. The dissertation provides a comprehensive analysis of the vulnerabilities in biometric recognition systems, particularly those related to the information stored in biometric templates. It demonstrates that fingerprint images can be recovered from Minutiae Cylinder Codes (MCC), a well-known fingerprint representation, with high accuracy, which can then be used to create spoof fingers and compromise enrolled systems. The dissertation categorizes techniques to protect biometric templates into two main groups: biometric cryptosystems and template transformation techniques. Biometric cryptosystems involve binding a secure key to biometric data to create a secure sketch from which no information about the biometric data or the key can be recovered. Template transformation techniques, on the other hand, non-invertibly transform the biometric template with the user's password. The dissertation studies two main examples of biometric cryptosystems: fuzzy vault and fuzzy commitment, and provides an improved security analysis that accounts for the non-uniform distribution of biometric features. It also proposes a framework to effectively combine multiple biometric representations. The dissertation identifies two limitations of typical biometric cryptosystems: linkability, which allows identifying two secure biometric templates generated from the same biometric, and the use of only simple biometric representations. It develops techniques to overcome these limitations in the context of fuzzy vault. Additionally, it evaluates various template transformation techniques proposed in the literature, assessing their security using comprehensive metrics. The analysis of template inversion, or the recovery of the original template from a transformed template, is a key element of the security analysis. The dissertation emphasizes the critical importance of protecting biometric templates for public acceptability and user privacy. It believes that the security analysis presented will help streamline the development of new techniques and find robust solutions for protecting biometric data.