초록(영문)
Nowadays, traffic congestion iscommon because the number of vehicles is mounting, while roads are limited. So, we felt a need of a new system which controls intersections efficiently by recognizing vehicles which are waiting for signal and pedestrians. It will shorten waiting time for both vehicles and pedestrians. In this research we pay attention to the raspberrypi. And we made real-time recognition system using raspberrypi camera module. Also, we checked the optimal height of camera to recognize vehicles effectively. After that, we designed two algorithms called 'Best Fit' and 'Multi Overlap' to control an intersection which uses data about the number of waiting vehicles and pedestrians. Finally, we confirmed whether our algorithms are effective or ineffective through simulator which we developed ourselves by using Game Maker. Before we simulate, we made four seeds which represent four different traffic situations. We performed experiments to check four things: the average waiting time of vehicles and pedestrians, the number of passed people, the time that simulator take to shut down, the number of signal transduction. As a result, 'Best Fit' and 'Multi Overlap' showed better performance than existing intersection-controlling algorithms in all respects.