Investigating the PCA Effect on the 3D Face Recognition System Speed

In this paper, we discuss increasing the speed of facial recognition using 3-dimensional data. The use of this data makes facial recognition safer because it is difficult to imitate compared to using 2-dimensional data. 2-dimensional data can be obtained from photos, but 3-dimensional data cannot be done because it requires a third dimension, namely facial depth data which can only be taken from 3-dimensional objects. The three-dimensional (3D) face recognition process in this study did not go through the stages of the facial reconstruction process into 3D form, but directly used data taken from the Kinect Xbox camera system, unlike most previous research studies. So that it can increase the speed and is not expected to reduce the level of accuracy. Data from the camera is directly inputted into the backpropagation algorithm. This algorithm was chosen because it is simpler than CNN or other algorithms, and also this research is only to prove the method used can increase the speed of facial recognition. In addition to eliminating the reconstruction stage, the PCA method is also used. PCA has a function to simplify the amount of data, thereby reducing the amount of computing which can increase system speed. Testing is done in two ways. The first test uses a combination of Backpropagation and PCA, while the second test uses only Backpropagation. The results showed that the system using a combination of Backpropagation and PCA achieved a speed increase of up to 34.2 times, but the accuracy was reduced by 8.5%. Further research is needed to answer the results of this study.