FaceNorm - Normalization Plug-in for Improved Face Recognition of Non-Cooperative Individuals

Carnegie Mellon University seeks $500,000 over a twelve-month period to develop the FaceNorm system which integrates image enhancement and normalization techniques into a plug-in module that can be inserted between existing CCTV cameras and face recognition systems to improve the recognition performance of any face recognizer. This work will extend, integrate and evaluate our current technologies for: (1) pose normalization; (2) expression normalization; (3) illumination normalization and (4) resolution enhancement. The goal of the project is to extend and integrate these techniques into a single module that can collectively cope with unconstrained pose, expression, illumination and resolution at the same time. The resulting FacNorm system will run in real-time on standard off-the-shelf PC hardware. ca/ncf