Ear Detection For Android Implementation And The Research For Ear Recognition

As one of the biological features,ear has the characteristics of universality,stability and easy collection,which attract researchers to focus on its rec.ognition.This thesis studies Android implementation of ear detection and the ear recognition algorithm.The main contents and results are as follows:(1)Based on the Haar-like feature,AdaBoost algorithm is used for ear detection.Th.e ear classifier under laboratory conditions and uncontrolled conditions is trained,respectively.The positive samples under laboratory conditions use the US IB-1 database,and the positive samples under uncontrolled conditions use the AWE database.The negative samples are all from the image segmentation database of the weizmann team.The experimental results show that this method has a high detection rate within 20ms under both conditions.(2)The ear biom,etric recognition algorithms are investigate.d.The HOG+PCA+SVM algorithm is firstly applied to the human ear recognition.Tn this method,the ear HOG features are extracted first,then PCA algorithms are used to reduce the feature dimension.Finally,.the recognition is conducted by a multi-SVM classifier.Comparing with the traditional PCA+SVM algorithm,the recognition rate of this method is higher than that of PCA+SVM algorithm,and the highest recognition rate can be 95%.(3)Ear detection is implemented on the Android platform.The human ear detection algorithm is transplanted to the Android platform through JNI technology,so that it can accurately detect the ear and focus on it.Results of three experimental tests demonstrate the effectiveness of the ear detection on the Android system.The ear detection realization in Android system lays a good foundation for the h,igh precision photography of 3D portrait modeling.Also an ear database of our own is established.The proposed new algorithm for ear recognition is a pre-study for further portrait recognition.