Abstract:
Currently the research of face recognition mostly aim on little internal face of simple environment and carried out on single computer, which limited the application of face recognition. The emergence of big data technology make parallel compute more easy, which make big scale compute work could be accomplished in very short time. An dynamic face recognition system was put forward in this paper. The system was suited for comparing little internal face with numerous face which were collected by large amount of camera. Storm flow compute model was used in the face recognition which could get recognition result in real time and will send the result to the people who interested in. Experiment shows that storm could get face recognition result in one minute regardless the number of internal faces and collected faces. When the internal face increase or the collected face increase, we only increase the servers in the cluster, the recognition time could be limited in 1 minute. Experiment also shows that because big data technology use parallel compute technology, which make more complex recognition algorithm, more complex feature extract algorithm, higher resolution image, and more internal image, and thus improve the recognition accuracy. Because the above could be fulfilled under parallel compute technology, the result could be gotten in real time.