Geometric ear recognitionbased on neural network

الملخص EN

-In this paper, an approach to recognize human using ear image biometrics using its features has been proposed.

Traditional image-based approaches to person identification almost exclusively employ front views of the individual's face.

Ear on the other hand, has a more uniform distribution of color, so almost all information are conserved when converting the original color image into gray scales.

The human ear is a perfect source of data for passive person identification in many applications.

In the growing need for security in various public places, ear biometrics seems to be a good solution, since ears are visible and their images can be easily taken, even without the examined person’s knowledge.

When employing face or lip as a biometric, changing of their appearance with the expression of the subject creates problem but in case of ear the shape and the appearance are fixed.

We extract the features an ear image by measuring geometric relations between predetermined points (lengths, angles).

Here, the geometrical structures observed from pixel value distances are used for the successful recognition of objects.

Recognition of objects is based on these geometrical structures.

Experimental result reveals that the proposed identification method can achieve almost 95% accuracy.