A flexible object model for encoding and matching human faces