Exemplar-based Face Detection

As a key technology of face information processing, face detection has become a popular research direction, which has academic value and commercial application value. Face detection is widely used in the fields of identity recognition, human-computer interaction, video surveillance, content-based retrieval, virtual reality and so on.A lot of achievements have been made in the research of face detection. In this paper, the current situation and development of face detection technology is summarized, which is based on the research of face detection in recent years. This paper presents exemplar-based face detection algorithm. In this paper, the exemplars refer to the face images marked facial landmarks and non face images marked random landmarks. The work done in this paper mainly include the following aspects:(1) A new shape-index feature is studied to cope with the effects of pose and facial expression. This feature is invariant to changes in the geometry of the human face, which can be a good way to eliminate the influence caused by the change of pose and facial expression;(2)A face alignment method based on random forest is studied. Shape-index feature is applied to the training of random forest. The information output of the leaf nodes in the training samples is combined into local binary feature, which can make a global regression for the offsets of facial landmarks;(3) Based on the face alignment algorithm of random forest, the cascade model of classification and regression tree is studied. Face detection and face alignment are combined into the same cascade model, and the shape-index feature is used to accomplish the classification and regression tasks.