Face authentication and non-standard image geometries: Issues, results and prospects

The analysis of biological features for the recognition of humans attracted the interest of many researchers in the last decades. Despite the considerable efforts, in many cases, the dimensionality of the data and the complexity of the matching problem made it difficult to devise fully operational authentication systems. This is probably also the reason of the greatest diffusion of other biometric modalities such as fingerprint and iris scanning. As in other applications, the adoption of non-standard image sampling geometries may both reduce the dimensionality of the data and the complexity of the recognition problem. In this paper we will try to demonstrate that this is true also in the case of face image analysis where non-standard imaging geometries can be successfully applied.