Construction of Frontal Face from Side-view Images using Face Mosaicing

In this paper, a method for fully automatic construction of frontal 2D face image of an individual is described. This frontal image is constructed from his/her available side-view images using face mosaicing. Firstly an automatic eye-brow detection algorithm is used to detect the eye-brow along with its position i.e. left or right. By this result it is easier to detect whether the given image is leftor rightside view of a face. After that an automatic face cutting algorithm is used to extract the facial areas and then alignment mechanism using eye-brow leveling is used to mosaic them to construct the frontal face image. Lastly eyes are shifted using geometrical transformation to make it almost perfect frontal view of the face. Also an approximation of the frontal face from a single side-view image can be generated by cutting it according to the previous method and copying the same half into the second half. Experimental results show that frontal face images may be easily constructed and those images can be used by simple face recognition techniques, which performs with low computational effort. Index Terms —Eye-brow detection, face alignment, eye shift,