A Proposed Template Image Matching Algorithm for Face Recognition

In this work, a suggested algorithm is developed based on normalized cross correlation algorithm in order to find the similarity measure between verified face images. Automatic matching of people faces is a demanding issue which has recently received increased attention during the recent ages due to its various uses in different applications such as law implementation, security requirements and video indexing. Face images are exposed to a variety of important influences such as illumination variation, difference in looking view, facial expression, occlusion, and age difference between compared face images and some individual changes in field settings. 

Most current face matching algorithms can be classified into two categories. Either geometry feature based or template image based. Normalized correlation is considered one of the methods based on template matching that can be used for finding a presence of a pattern or a feature within an image. Here, a Graphical User Interface (GUI) is prepared to perform all face matching tasks and is used to test face verification. This GUI has been examined by conducting many case studies. The concluded results indicate that the developed algorithm is so robust in face matching and shows good performance.