Multimodal recognition based on pose transformation of ear and face images

Ear and face recognition have possessed of important stations in biometric field. Pose problem, however, has always been a challenge for researchers. Therefore this paper proposes a novel method based on pose transformation to solve pose problem. In addition, the generated ear and face features are fused by serial strategy parallel strategy and KCCA, and are identified by the nearest neighbor method. Experimental results show that it is effective remarkably for monomodal biometrics including ear and face to use pose transformation, and multimodal biometrics outperforms monomodal ones.