3D FACE RECOGNITION BASED ON DECISION FUSION

In this paper, we propose the algorithm of 3D face recognition based on decision fusion. In the algorithm, firstly, it uses PCA or LPP to extract the features of 2D normalized grey image and compute Euclidean distance between the testing sample and the training sample as one matching score. We process the 3D depth image with multilevel B-splines approximation and ICP method correction. Then we can obtain another matching score by using PCA or LPP to extract the features of 3D depth image. Finally, decision fusion results can be gained by fusing the two scores. The experimental results indicate that the recognition rate of the decision fusion method which uses LPP to extract the 3D image features is higher than any single algorithm, and the recognition performance is improved greatly.