Recognition face in picture by means of analysis of special regions

A new approach to face recognition problem is suggested. The approach is based on the hypothesis that human beings recognize person in picture by means of an analysis of brightest spots in picture. Comparing two pictures of the same person one has to level sizes of the pictures. To this end a special procedure that was developed by the authors earlier is implemented. After the leveling is completed the brightest regions in the picture are detected by the technique of non-orthogonal wavelet transform. Each transformed picture is replaced by its Fourier transform. The final decision based on a compare of the found Fourier coefficients.