混合动态纹理与李群论相结合的交通异常事件监测技术研究

交通流视频背景较为复杂，当前交通异常事件监测技术无法有效抑制复杂背景下的噪声干扰，导致监测精度低下。为此，提出一种新的混合动态纹理与李群论相结合的交通异常事件监测技术，通过Gabor滤波器对交通视频图像进行滤波处理，采集图像纹理信息，通过时空方向能量对交通异常事件图像动态纹理特征进行提取。依据李群空间中仿射群组不受外界干扰能够保持形状的特性，把交通异常事件状态变量映射至李群空间进行处理，介绍了李群和李代数之间的空间映射关系，给出交通异常事件状态方程和相似匹配度量准则，利用跟踪图像的加权均值对跟踪模板进行更新，实现交通异常事件的监测。实验结果表明，所提技术监测时效和精度均较高。

Hybrid dynamic texture combined with the theory of lie groups of abnormal events traffic monitoring technology research

The background of traffic flow video is complex, and the current traffic anomaly monitoring technology can not effectively suppress the noise interference in the complex background, resulting in low monitoring accuracy. To this end, the traffic incident monitoring technology of a new hybrid dynamic texture and Lie group theory combined, through the filtering of the traffic video image Gabor filter, image texture information, based on the traffic incident dynamic image texture feature of temporal direction of energy extraction. On the basis of the affine Li Qun space group without outside interference characteristics to maintain the shape of the traffic incident state variables mapped to the Li Qun space, the space mapping relationship between Li Qun and the lie algebra, given the traffic incident state equation and similarity matching measurement criteria, the tracking template is updated using the weighted average image tracking the implementation of monitoring of traffic incident. The experimental results show that the proposed technique has high accuracy and efficiency.