摘要
自行车具有骑行速度慢、行驶轨迹多变等特点,传统检测手段难以获取车行轨迹数据。提出基于视频检测来研究自行车流跟驰特性的方法,在对视频检测交通流所获得跟驰事件定性分析的基础上进一步提出用K-means聚类定量判定出跟驰样本,并根据筛选出的样本建立自行车流的广义回归跟驰模型,统计分析结果表明该方法是有效的。
Bicycle riding behavior is slow, small and track changing irregularly, the traditional traffic detection methods are difficult to obtain track data, the paper proposes to study the behavior of bicyclists in following based on video detection. On the basis of bike following event qualitative analysis put forward the quantitative method using K-means clustering to determinate the bike-following samples, and eventually developed generalized regression bicycle flow following model. The result of statistical value indicates the method’s effectiveness.
出处
《交通科技与经济》
2010年第5期43-45,共3页
Technology & Economy in Areas of Communications
关键词
自行车流
视频检测
K-MEANS聚类
跟驰模型
bicycle traffic flow
video detection
K-means clustering
bicycle flow following model
