Symmetry sensing through computer vision and a facial image recognition

The symmetries exhibited by nearly front view photographs of a human face have been explored by using group theory. Thus we develop an algorithm for facial image recognition implemented through computerised symmetry sensing for dealing with problems of personal identification from face photographs. The questioned facial image is made superimposable over the control one by processing through a properly computed transformation of projective symmetry. A total superimposed image, as well as other composite images produced from the control and the processed questioned images, facilitate accurate decision-making in identity establishment. All such images can be stored as permanent records and their hard copies or photographs from the monitor screen are suitable for court demonstration. An example of personal identification in a case of suspected impersonation has been demonstrated as a practical application of this new method.