期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2013
卷号：52
期号：1
出版社：Journal of Theoretical and Applied
摘要：Running red lights (RRL) currently become a hotspot social issue. Although it is commonly mentioned to detect and warn by using traffic monitoring, this idea has still remain in human recognition. Therefore, this paper designs a method to remedy it, which can detect the behavior of the pedestrians crossing the road automatically. From the sequential images taken by traffic monitoring, the use of optical flow estimating the track of motion objects whether cross the designated area enables automatic identification, then we carry out frames subtraction to isolate and identify the pedestrians from the moving objects. Combine the advantage of both algorithms, a system is presented to recognize the RRL event. Ultimately, the experiment confirms and indicates that what we proposed works effectively in some way.
