����Kalman�˲������˸��ٷ����о�

应用视频处理技术对行人交通进行研究和分析受到广泛的重视和发展，已成为智能交通领域的研究热点之一，而行人跟踪是行人交通采集系统的基础，是后续交通参数提取，行为分析的前提.本文提出一种基于Kalman滤波的行人跟踪方法.首先，利用改进的混合高斯模型提取背景，并采用HSV颜色空间模型和目标重构方法得到消除阴影后的前景图像.其次，进行行人的特征提取，得到融合行人位置特征和形状特征的运动模板.最后，采用Kalman滤波对行人的运动轨迹进行预测，将检测到的目标与预测结果匹配，得到行人的跟踪匹配矩阵,根据匹配矩阵判断合并和分离的发生.试验结果表明，该方法不仅可以准确跟踪多目标，而且可以有效解决遮挡问题，具有很好的适应性和鲁棒性.

Study on Pedestrian Tracking Based on Kalman Filter

Pedestrian tracking is an important part of Intelligent Transportation Systems (ITS) application. A new robust system for pedestrian tracking based on computer vision is proposed in this paper. First,improved Ganseianmixture model is used to detect pedestrian. To get better segmentation result,morphological reconstruction is utilized to remove shadows. Then,pedestrian tracking is achieved by feature fusion and prediction methodology. The trajectories are incorporated by the Kalman filter to determine the search windows,which are used to establish thematching matrix at two successive frames by comparing with detected pedestrians. Finally,an object association strategy is incorporated to deal with object tracking in case of merging and splitting by the template integrated spatial position and shape information. The tracking method is tested under real scenarios. Elaborate experiment results show good robustness and high efficiency of this method.