摘要
在城市交通流量视频检测系统中,昼夜转换是必须面对的问题,在白天和黑夜的过渡期间,简单的使用白天算法或夜间算法检测效果较差。本文提出一种针对昼夜转换场景的车辆检测算法,该算法首先提取出背景图像,针对昼夜转换场景中光线昏暗、变化较快的特点,建立了一种能够快速跟踪背景变化的背景更新模型;然后采用背景差分的方法检测运动车辆。实验表明,本文算法能够很好的检测昼夜转换场景中的运动车辆。
In the video-based urban traffic flow detection system, day-night shift is an important problem, Whatever the day or the night algorithms are not suitable for the day-night shift scenes, and detection result are unsatisfied. This paper presents a vehicle detection algorithm for day - night shift scenes. In the day-night shift scenes, the environment is dim and light intensity changes quickly. So the algorithm firstly builds a background update model that can quickly follow up the background shift. After that, the background subtraction method is applied to detect moving vehicles. The experiment results show that the algorithm can accurately detect the vehicles in the day - night shift scenes.
出处
《信号处理》
CSCD
北大核心
2006年第3期390-394,共5页
Journal of Signal Processing
关键词
昼夜转换
背景更新
车辆检测
Day-night shift
Background update
Vehicle detection
