3D Face Recognition using PCA and Convolutional Neural Networks
Nowadays, face recognition which is a branch of artificial intelligence, is continuously implemented in many fields, therefore, a more accurate and reliable system is needed. Currently, the 2D face recognition system is commonly used for system security, but has weakness that system is able opened by using a photo. To overcome this problem, there is need for the development of a 3D face recognition. This system is more secure because has data depth that is not owned by photo. To produce 3D data, Tof camera is used. Kinect Xbox One is one type of such cameras which is dependable and relatively cheap. In previous research, the 3D face recognition system was built using Kinect Xbox One, while the PCA and ANN algorithm was used for its implementation. However, the accuracy of the system was approximately 80%, therefore, the CNN, which is a more sophisticated algorithm was used for enhancement accuracy. The image size obtained from the Kinect Xbox One was reduced by the PCA algorithm before conducting it to CNN and from the test result, the accuracy level is increase become 90%. Therefore, the PCA and CNN implementation at the 3D face recognition system proved to increase the accuracy system.
Keywords - Biometric, Face recognition, Face 3D, Tof, Kinect Xbox One.