People recognition through biometric identifiers has a variety of applications today. This process is performed in plain text, which endangers the safety of data transmitted during the recognition process when performed in low security networks. Biometric identifiers are unique and invariant in the lifetime of an individual. Therefore, once the data associated with the biometric identifier are obtained, it cannot be safely use as a security mechanism, so is not possible to change it as a password or a personal identification number. For the cryptographic protection of biometric data, several models have been proposed, however, problems such as data alignment and the revocation of compromised templates during an attack have not been efficiently addressed in these models. A set of hybrid models have been proposed in the literature to facilitate the revocation and to introduce a contribution to remove the alignment process. For this end, several minutiae structures extraction mechanisms are involved, with the purpose of obtaining a model that uses a method for extracting information invariant to rotation and translation, resistant to nonlinear deformation and partial overlapping and one of the biometric cryptosystems to ensure the extracted set of data.