3D Ear Reconstruction and Identification Method Research Based on Morphable Model

3D Ear recognition has become a new subject in the biometric authentication field. In this paper, ear recognition using three-dimensional ear model reconstructed by morphable model method is studied. In the morphable model construction phase the 2D Procrustes analysis method is used to correct posture. Sparse model based matching methods is applied to adjust model parameter using 2D image. After the reconstruction of 3D ear models, the Euclidean distance between the reconstructed model and the known model of the UND database is calculated as the modeling quality assessment. For the identification experiment, morphable model is used to reconstruct the 3D models of UND and USTB sub ear database. Then the recognition experiment is conducted. The experiment results show that the application of 2D Procrustes analysis in posture correction is effective, and also prove the validity of recognition using three-dimensional ear model reconstructed by morphable model method.