A pose and expression face recognition method using transformation based on single face neutral reference

Abstract

To recognize a frontal face image with pose and expression is one of challenging research problems, which are often failed during matching with neutral face database. In this paper, we propose a pose and expression face recognition method by converting a given face image into a single face neutral reference to create a synthesized frontal face image that is more similar face with the database. We, thus, transform a pose face to a frontal face and an expression face to opened-eyes and closed-mouth face by using triangle transformation. Pose views of face image used are yawing and pitching view that may contain varied expressions such as neutral, disgust, scream, smile, squint, and surprise. We evaluate our proposed method by using well-known face database; namely, CMU Multi-PIE database and CMU PIE database. Our proposed method improves the face recognition accuracy when using our synthesized face image via virtual neutral face database.