摘要：Study on bus passenger recognition and track of video sequence is a research aiming at realizing making a statistics on passenger flow volume of bus, which not only helps traffic services center rationally schedule vehicles, but also can avoid overload. The system makes machine vision technology and digital image processing technique applied in customer counting of bus. The key questions in the study include effective collection of video image, effective extraction of moving object in sequence images, recognition of moving targets, trace and count of moving objects. The paper makes deep analysis on relevant technology from the perspective of the theory and practical application of the algorithm, and the study makes achievements. The theoretical algorithm for the study on bus passenger recognition and track of video sequence is distinctive. The practical experiments indicate that the system can satisfy the real-time requirements, and can accurately make statistics on the number of passengers, which has great application value.
关键词：machine vision;frame difference;fuzzy clustering;Kalman filter;cost function
