With the rapid development of economy, the person flow pressure for the station, market, subway and other big public places is very largely to current society. The counting and monitoring of person flow is very important in society management and development, it has great practical significance in the public traffic system safety monitoring, passenger flow forecast, the optimal allocation of resources etc. Traditional person flow estimation method takes the manual counting method, and the efficiency and accuracy are bad. An improved person flow counting method is proposed based on video image processing and on feature point extraction of optical flow, and the person flow monitoring system is designed based on the algorithm, the video image is processed with Harris corner detection, and the angle point information of image is obtained, according to the continuity of the optical flow, the flow counting is obtained. The system design process is presented based on the feature point optical flow. Simulation result shows that the system has good performance in person flow counting estimation and monitoring.