The Research And Implementation Of Passenger Automatic Recognition System In Railway Passenger Transport Hub

Passenger automatic recognition has important practical value and broad prospects in passenger transport, shopping malls and public security area. And at present passenger automatic recognition is a hot research issue in China, including various fields and many core issues, such as video image acquisition, motion object detection, and image processing and so on. This thesis is a part of National 863 Project, "Pedestrian Safety State Recognition Technology of Distributed Service Network in High-speed Railway Integrated Passenger Transport Hub", and focuses on the research and implementation of passenger automatic recognition system in railway passenger transport hub.In the thesis, at first, according to the characteristics of railway passenger transport hub and development trends of video image acquisition, video image acquisition system is designed. The system is based on streaming media network, making the passenger automatic recognition system realize its images collection remotely, without affecting the operations of other equipments in the whole video image acquisition system.The core algorithm and innovation part of the passenger automatic recognition system is the passenger automatic recognition model of optical flow based on background modeling. First, use single Gaussian background modeling to abstract background, and then get the foreground by background subtraction. Calculate optical flow of foreground to detect moving targets and divide overlapping targets. At last use image processing algorithm to optimize, and get passenger numbers by connected domains analysis. This model decreases iterations, improves real-time calculations, and provides high accuracy and practicability.Finally, use mixing programming of C# and MATLAB to realize modules of the passenger automatic recognition system in railway passenger transport hub. After the system test and results analysis, the system shows high practicability.