摘要
车流量检测是智能交通系统中的关键技术之一。研究了多种基于视频图像处理的车流量检测算法,包括基于灰度图像的背景差分法、帧差法、边缘检测法和基于彩色图像的色彩跳变检测法。在分析了以上算法在不同检测环境中适用性差异的基础上,提出了1种修正的背景差分法,并在此基础上实现了1种通用性更强的综合检测法。综合检测法结合背景差分法,边缘检测法和色彩跳变法三者优点,可依据光线条件自动选择检测区域和检测算法,适用于多种检测环境,准确率超过90%。
Vehicle flow detection is a key technique of intelligent traffic system. Many video-based vehicle flow detection algorithms are discussed in this paper, including gray level based hackgronnd difference, frame difference, edge detection, and color-based hopping detection. First, the performance difference of these algorithms in various environments is analyzed. Then, a modified algorithm of background difference is proposed and a synthetic algorithm with broader applicability is performed. In this synthetic algorithm, the advantages of background difference, edge detection and color based hopping detection are put together with some improvements and both the detecting algorithm and detecting arWith over 900% accuracy, this algorithm is more feasible than current ones.
出处
《交通信息与安全》
2010年第1期20-25,共6页
Journal of Transport Information and Safety
基金
国家自然科学基金项目(批准号:30970780)资助
关键词
智能交通系统
车流量
视频检测
intelligent traffic system
vehicle flow
video-based detection
