Face Recognition and Tracking System Based on Embedded Platform

Abstract

The paper introduces a face recognition and tracking system based on My RIO (National Instruments, USA), LabVIEW, NI Vision tool kit and OpenCV library. The system includes two parts, static recognition and dynamic tracking of face detection. The static part is mainly based on the combination of face feature extraction and template matching to realize the function of face recognition. The dynamic part is based on the combination of Haar classification and Camshift algorithm, through which the system completes the task of tracking face. The results show that the system works optimal when the threshold of matching is set as 60. To a certain extent, the accuracy of the system is affected by the illumination. With poor lighting, the system's recognition rate can still reach 72.10/0 and the rate of tracking gets to 83.4%. From the above performance, the system works well. Therefore, the system is significant to authentication and other fields.