【摘要】 为了提取高密度混合交通流下的自行车交通参数,建立基于视频的交通检测系统,改进了卡尔曼自适应背景模型,提出了基于决策树的自行车群识别方法和基于面积阈值的车辆计数方法。对实际场景拍摄的视频处理得到了混合交通流分类、计数结果和速度密度关系。结果表明,该方法不但能有效地检测、跟踪、识别高密度混合交通流下的自行车目标,而且达到了为自行车模型研究获取合理的交通流密度、速度等相关交通参数的目的。更多还原

【Abstract】 To fetch the parameters of bicycles in heavy multi-traffic scenes, this paper developed a video traffic monitoring system firstly, then improved Kalman filtering based adaptive background model, and presented a new bicycles recognition method using decision tree and a counting algorithm based on area threshold respectively. Extracted multi-traffic classification, counting results and velocity-density relations from the actually scene video. The results indicate that the system can detect, track, and recognize the bicycles and can get the reasonable traffic parameters like density, velocity, etc.更多还原