基于高斯混合模型和卡尔曼滤波的车辆检测与跟踪方法

提出一种改进的车辆检测与跟踪方法。在目标检测阶段,针对传统高斯混合建模算法对环境变化适应能力较差的问题,设计一个环境变化判断因子,据此进行高斯混合模型更新率的自动切换;在车辆跟踪阶段,为提高跟踪精度和跟踪效率,引入卡尔曼滤波并设计了跟踪列表进行单目标和多目标的跟踪。实验表明,该方法对光照突变有较好的适应性,能实现车辆的有效检测与跟踪。

Vehicle detecting and tracking method based on Gaussian mixture model and Kalman filtering

An improved vehicle detecting and tracking method is proposed. In the target detecting stage, aiming at the problem that traditional Gaussian mixture model (GMM) method has weak adaptability to environmental changes, a judgment factor about environmental change is designed and used in the automatic switching of the update rate of GMM. In the vehicle tracking stage, Kalman filtering is introduced to increase the tracking accuracy and efficiency, and a tracking list is designed for single and multiple target tracking. Experimental results show that the proposed method has strong adaptability to the abrupt change of illumination and is efficient in vehicle detecting and tracking.