International Journal of Engineering Associates
Face recognition systems are being used in several applications, such as access control, surveillance systems, and forensic analysis amongst others. In non intrusive systems, the subject to be identified may not be in the appropriate position at the time of authentication. The face image obtained may be subject to certain transformations, thus, there is a need for the system to cater for such flexibility. This paper investigates how a normalization process can cater for recognition of face images that have undergone transformations namely translation, scaling and rotation. The method used here first detects the region of interest, the face in the transformed image and tries to convert it back to its original form.