Abstract: In view of shortcomings of low recognition rate and even unable to identify coal mine personnel existed in traditional face recognition algorithm, the paper proposed a face recognition algorithm based on SURF and FLANN. Firstly, the algorithm uses fast-Hessian detection matrix of SURF algorithm to find features, and generates feature vector of SURF descriptor. Then it adoptes search algorithm of positive and negative characteristics of Hessian detection matrix and bothway FLANN matching algorithm to match SURF descriptor, so as to realize face recognition and check on worker attendance．The experimental result shows that the algorithm can not only exclude mistake matching points，but also improve recognition speed and correct rate of SURF algorithm ,and ensure the real-time performance of the algorithm in the attendance system.