Automatic Feature Detection Using Pattern Projection for Person Identification

In this paper an efficient algorithm is proposed for person identification based on the analysis of 2D images of a designed mask projected onto the person's face in a cooperative environment. The cooperative environment is essential for the accomplishment of the images under a certain condition. The importance of the designed mask is for avoiding the complexity of feature selection from the face image. The mask contains a number of designed patterns with single center, minor and major axis. Due to the structure differences of human faces in 3D, 2D images of the proje~ted mask contain different information about the shape of an individual face. Each boundary of a pattern on the original mask is considered as an individual object's shape on the image plane. For an image plane each shape is matched with the same number of pattern in the image. Simulations are conducted with real data to confirm the effectiveness of our method.