Security is a critical concern for any database. Therefore, database systems provide a wide range of mechanisms to enforce security constraints. These mechanisms can be used to implement part of the security policies requested of an organization. Nevertheless, security requirements are not static, and thus, implemented policies must be changed and reviewed. As a first step, this requires to discover the actual security constraints being enforced by the database and to represent them at an appropriate abstraction level to enable their understanding and reenginering by security experts. Unfortunately, despite the existence of a number of techniques for database reverse engineering, security aspects are ignored during the process. This paper aims to cover this gap by presenting a security metamodel and reverse engineering process that helps security experts to visualize and manipulate security policies in a vendor-independent manner.