Research On Some Issues Of Ear Recognition

Being a mature identity recognition, biometrics has already applied in many fields, such as face recognition, fingerprint recognition, iris recognition, sound recognition etc. Ear recognition is a novel technique in the field of biometrics. Ear recognition can be a beneficial supply for biometrics recognition, because of its unique physiology character and the observation angle. At the present, ear recognition is underway in domestic and abroad, so there are many basic problems that are not solved. More and more experts pay attention to its uniqueness, practicability and theory research value.According to the relationship between ear contour and proportion of the five sense organs, this paper adopts a method for biometrics based on combination of ear contour's geometry parameter and facial configuration feature. This method combines some features of face and ears, then forms the higher-dimension feature vector. The feature vector consists of the proportion of ear's compared width with pupils' horizontal distance, the proportion of comparing ear's height with vertical distance between eyes and mouth, the proportion of comparing face's width with pupils' horizontal distance etc, eight features in total.Firstly, we should remove hair in binary image, and then locate eyes' position with the method of gray projection and area growth. Secondly, we adjust the image's position, in order to gain the correct face image. We use template to locate mouth, then trace ear's contour. There is a vector according to the relation of structure of eyes, mouth, face and ear contour. The eight-dimension features have different effect in pattern classification process due to different characters. The eight-dimension features have different effects in pattern classification process due to different characters. In the process of pattern recognition, we use Euclidean distance formula to classify sample.The features that are relative values reflect ear's shape and size, the angle between ears and head, the vertical distance between eyes and mouth, the distance between mouth and ears etc. So it can avoid the deviation from the camera and target. If we remove the ear's features, only by face features can reduce the difference among individual. The experiment's result shows that ear's features enhance the vector's performance. The right recognition rate is 86 present in the ear recognition system.In the summarize section, there are conclusions about all work previous and expectation about research in future.