Research On Ear Characteristic Vectors Based On Ear Structure

Ear recognition is a new characteristic biometrics which follows face recognition, finger print recognition and so on. It pertains to the use of an individual's ear form to determine or validate identity. However, there are few research in this new field. So the ear recognition should have enormous potential and large room of application and development.Since the ear recognition belongs to recognition field, it must depend on traditional recognition technology and methods, especially face recognition and finger print recognition. On the other hand, as a result of the ear's physical position and shape, there should be a particular recognition method with it. Here this paper mainly studies the pretreatment of the ear and then extract some characteristic vectors from the ear image.At present, the ear recognition is not a mature technology. There are many problems in this new field to solve. In this paper, we mainly use manual image division, but it is not accurate enough to achieve the automatic ear distinguishes request. This paper carried on the preliminary exploration in this aspect, concretely speaking, the ear image was transformed to binary image according to the image segmentation by thresholding of the ear image. After we get the obvious contour of the ear, some geometrical characteristic vectors also can be obtained, including contour curve polynomial, inside angle of ear , scale of contour's length and ear axes.This paper mainly tried some beneficial research and study in the ear recognition, which will do some good to the development of the ear recognition in some extent.