The Research Of Statistical Technology For Bus Pedestrian Flow Based On The Image Processing

At present, with the rapid development of China’s economic society and the continuouscreasing in urban population, the public transit system, as main traffic tools of residents, willcome under more and more pressure. Therefore, the intelligent transportation system based oncomputer vision has become an increasingly important role in the urban traffic management.The method for counting the passenger flow by using image processing technology has beenone of the hot-points in this research field. Through statistics of bus passenger flow data, thispaper analyses urban residents travel OD. It is great significance to develop operationalsummary operations plan in the public transport management department, in order to meet theneeds of taking buses in different period.At first, this article analyzes significance of the passengers flow statistical system,summarizes current statistics methods of public transport passenger flow. Then studies basicknowledge of image processing that will be used in this paper. This paper analyzes theadvantages and disadvantages of several motion detection algorithms, and presents themoving target detection that is more adapt to the environment of public transport based on theedge of the test. Combining with the real effect of video, the paper using Hough transform forfeature extraction of passengers head, achieves good results, which laid a good foundation forlater target track. Finally, we take Camsift algorithm based on the Kalman prediction fortarget tracking, and count the passenger according to the tracking information. In this paper,the image is filmed under the simulated buses condition of laboratory, and the passengers areshot by using the top mounted camera. The algorithm in this paper is finished by usingOpenCV library on Windows operating system.