Compact multi-level representation of human faces for identification

In the last 15 years the task of face recognition has attracted considerable research attention. A number of advancements in automated face recognition use a variety of approaches. Although every technique has a particular merit, it is not yet possible to achieve fully reliable recognition, even in constrained environments. Specific problems that need to be solved are 3-D rotations, expression invariance, different appearances of the same person, integration of information from different scales, etc. The goal of this research is to solve the problem of integration of high-level information into a low-level representation of faces. Such integration results in a more efficient recognition performance. The subordinate advantage of this recognition scheme is a compact face representation, which reduces the database size and speeds up the recognition process.