Summary
The threat caused by criminality and terrorism resulted in a significant increase in the interest in research focused on effective methods for its reducing. One of the most important solutions for the mentioned problem is the biometric identification of persons. Hence, it became a very important issue nowadays. A crucial aspect of biometrics is the searching for automatic methods that may be applied for the recognition of human beings. Currently, fingerprints as well as the face are the most popular applied biometric features. However, in order to increase the efficiency of the developed systems, new modalities are becoming more and more popular. On one hand, researchers are looking for more effective biometrics while on the other hand the idea of applying few modalities jointly in multimodal systems is becoming more popular. An ear is an example of lately explored new biometric features. Its uniqueness is the most important advantage. Similarly to the face, the auricle is distinguishable for various persons, thanks to its complex and stable structure. Therefore, in the paper an algorithm for human identification based on ear images is presented and tested. It uses the improved version of the Polar-Fourier Greyscale Descriptor for feature representation. The method was tested using 225 digital ear images for 45 persons and has achieved 84% efficiency.