Suspect Face Retrieval via Multicriteria Decision Process

Using human definable face attributes, this paper presents suspect face identification approach for criminal investigations. To facilitate this approach, Fuzzy Analytic Hierarchy Process (FAHP) weights were created to produce the salience of facial attributes. Facial Attributes, such as short, medium, and long, are given manually assigned weights based on membership grades. The accuracy of face retrieval from a face repository is evaluated through experiments. Along with eliminating the need for expert artists, the proposed method complements sketch-based face recognition by enabling investigators to search face repositories immediately without having to wait for sketches to be generated first.