A geometric approach to face recognition.
В. В. Старовойтов, Dmitry Samal · 1999
We present our study of the face recognition problem. Due to difference in human pose, face expression, hairstyle, image style and lighting conditions, the problem is very difficult. To solve it we have to test different image processing tools and heuristics for robust recognition. The main features of our approach are detection of fiducial points, calculation of geometric features and application of nonlinear image dissimilarity function at the final recognition stage. We demonstrate the power of the approach by experiments. 1.