Pose Robust and Person Independent Facial Expressions Recognition Using AAM Selection

Most recent facial expressions recognition systems only work well with frontal face images. However, subjects do not always face front. With this in mind, we propose in this paper a method for pose-robust facial expressions recognition. Active Appearance Models (AAMs) are used for face tracking to extract pose-robust facial feature points. However, AAM has accuracy problems with face tracking when it tracks an unknown face. To solve this problem, a method was already proposed to construct plural AAMs by clustering the training datasets and then selecting one of their AAMs that is similar to the unknown input face based on the Mutual Subspace Method (MSM). in addition to that method, we constructed models based on face direction. The experimental results showed an improvement in the accuracy of facial expressions recognition.