Design And Development Of Bus Passager Counting System Based On Video

As an important part of intelligent bus dispatching system, bus passenger counting system based on video has important use value in the field of intelligent transport. At the same time, as a kind of incision type people counting system, bus passenger counting system based on video can be also applied to traffic analysis, crowd monitoring and any other fields, which has broad application prospects. In the basis of cooperative company’s existing monitoring system, this paper designs and develops a bus passenger counting system based video on embedded platform.This paper first explores techniques and methods used in this passenger counting system based on video by research of the existing techniques of people counting system based on video and comparison of moving target detection, target segmentation and motion tracking. Then this paper gives the overall design of this system by solving the problems that may occur in this system. Next this article introduces key algorithms of the system by target segmentation, target detection and tracking and people counting and presents a solution to the situation that statistical algorithm cannot be achieved under high traffic density.In the part of target detection, this system uses a popular algorithm at present named codebook. In the part of target segmentation and tracking algorithm, this paper proposed a target tracing method based on target overlap area in the basis of the target segmentation based on projection. This method is simple and efficient and suitable for tracking of slow moving targets.In the part of people counting, this paper brings up a counting method based on the target area. The algorithm can effectively shield the count disturbance that passengers reciprocating near the counting line. In the part of high density traffic counting algorithm, this paper proposes a recognition method based on passenger head color features. This method has the same algorithm with the target segmentation, tracking and counting. After that, the starting conditions of the high density traffic counting are given in this paper, which means the algorithm is started when passenger density reaches to a certain degree.This paper introduces the development process of this system, which include the system hardware and software platform, application development and the migration process. The OS of the embedded hardware platform based on Arm is Linux. Application on the development of this paper is a combination of OpenCV which is a popular visual library and Qt which is a cross-platform graphical interface application framework. In the end, this paper testes the system and reports the evaluation.In the basis of embed platform based on Arm, this paper implements an efficient and robust people counting system based on video. This system has strong scalability, solving the problem that statistical data of passengers cannot be obtained by the intelligent bus dispatching system in real time.