A FPGA implementation of facial feature extraction

Face recognition has a wide range of applications, such as biometric identity authentication, multimedia monitoring/surveillance, personal device logon and tracking. This paper describes the implementation of face recognition system using a Field Programmable Gate Array (FPGA) and a computer. Our proposed design uses the feature-based method to recognize a face. A computer is used to store the 17 relevant facial feature points for each face template per person. During the face recognition process, the FPGA first captures a video frame and then extracts the locations of 17 facial feature points from a face. These data are then transmitted to the computer through the serial data link (RS232). The computer then calculates dissimilarity scores for each face template stored in the database and finds the best match. We have measured the performance and tested the speed of the proposed system. The experimental results demonstrate that our system has the potential to be applied to dedicated devices such as an access control system.