摘要： 针对夜间环境下的车辆检测问题，从车头灯视角出发，提出一种具有高鲁棒性的夜间车辆定位和跟踪方法。结合卡尔曼滤波实现健壮的亮斑帧间跟踪，并根据亮斑的运动连续性和形态稳定性提取车灯目标。采用基于时域和空域特征的谱系聚类方法对车灯进行同车分组，利用车头灯组对车辆目标进行准确定位和跟踪。实验表明该方法在夜间交通环境中的有效性和高鲁棒性。

关键词: 智能交通系统, 夜间车辆检测, 卡尔曼滤波器, 谱系聚类, 定位与跟踪

Abstract: This paper proposes a high robustness method of nighttime vehicle location and tracking from the headlight angle of view. By virtual of Kalman filter, a robustness tracking process of each bright spot is realized and vehicle lamps are subtracted exactly according to their motion continuity and shape stability. Pedigree clustering methods based on the temporal and spatial features are brought in to classify the lamps into vehicle lamp groups. The headlight groups selected from the lamp groups are employed in the accurate location and tracking of moving vehicles. The validity and high robustness of this method is proved in the experiment.

Key words: intelligent transportation system, nighttime vehicle detection, Kalman filter, pedigree clustering, location and tracking

中图分类号: 

• TP391